/v/ - Video Games

it's fucking video games, baby

New Reply
Files Max 5 files32MB total
[New Reply]

Putin's given us the boot! Read about it here: https://zzzchan.xyz/news.html#66208b6a8fca3aefee4bf211


[Hide] (1016.1KB, 1517x1209) Reverse
>#8/agdg/ via irc.rizon.net
>https://matrix.to/#/+agdg:matrix.org via matrix programs
>Dev resources: http://8agdg.wikidot.com/resources
>Wiki: http://8agdg.wikidot.com/
>Previous: >>202641
Continuing from the previous thread
>Textures are procedural these days, you dont paint, you generate them.
That works if you want, say, a wooden table, or a rock, but what about more regular things made up of multiple materials. Like a speaker, a shipping container with various markings, and old aluminum can with a still visible label, a t-shirt, etc.
>just get completed ones for free
Tried that, but I couldn't find anything that looked anything what I wanted. And everything is PBR these days, which is fine if you want the asset-flip unity/unreal pseudo-realistic look, which is not exactly my goal.
Replies: >>220544
[Hide] (83.4KB, 948x735) Reverse
im going to (cheap) college to learn how 2 maek game
Replies: >>220438 >>220544
you won't learn shit there except how to copy and paste from StackOverflow and how to stuff as many ugly niggers into your game as possible. install godot and read their docs.
Replies: >>220460
might as well install gmod and pretend you made a game
Who will be the first zzzchan anon to start and finish game?
Replies: >>220538 >>220553
[Hide] (950KB, 1000x1000) Reverse
[Hide] (3MB, 1000x2000) Reverse
Than learn how to generate your own textures. 
>what about more regular things made up of multiple materials. Like a speaker, a shipping container with various markings, and old aluminum can with a still visible label, a t-shirt, etc
Procedural rust. 
Procedural edge wear.
>i want style
Find someone to work with.
Replies: >>220557
I already did but no one played it.
Post it and I will play it. Unless it's on steam because steam isn't on my OS.
Replies: >>220561
But I still not get it, like those nails in the last video. Am I supposed to create a different object/material for each little screw and whatever I have? But I want models that I can render real-time in a game engine, not a slideshow.
Replies: >>220559
Procedural materials are procedural, they are automatically generated. You make them, adjust them to look good, and than you Bake them into textures. 
This ^ video should explain things. 
You bake textures and retopologise the meshes. 
>different object/material for each little screw and whatever I have
Modern blender has free material libraries, where you can simply drag+drop whatever you want, but no, you dont need a separate material for each screw, a single material would probably suffice for all metal parts of an object. 
But its not easy and simple, its real work. Both proper remeshing/baking textures/texturing/modelling and learning how to do it. 
But alternatively you can use these "guides", you basically skip all of that and just do modelling+color painting from palette. 
Replies: >>220588 >>220683
Hey, I tried that when you posted that and I sucked
Is that you on the top left of the page? Hang in there
Well I played and although I think the vibe is great, the gameplay is shit. I got to the door on second level but was missing a key and couldn't be bothered to reset.
Replies: >>220567
[Hide] (214KB, 800x844) Reverse
Replies: >>220569
I want to FUCK clownpiece.
I will add her as a marriageable npc in my game.
Holy shit, you finished this?  Damn nigger, I never saw you announce it.  You need to shill yourself more.
Are you the anon who posted their RTS game on Steam?
I've hit a major road block with doing my 3d modeling in that I can't texture anything for shit. I could really use some help, but the major problem isn't just texturing it's making the textures themselves. Flat coloring everything just looks awful.
Replies: >>220602
[Hide] (1.1MB, 1024x1024) Reverse
[Hide] (231.7KB, 1582x2048) Reverse
[Hide] (110.1KB, 1024x768) Reverse
I still don't get it, let's suppose I have something like pic 1 or maybe pic 2. Am I supposed to generate all those screws and whatever procedurally?
Or as a more extreme example, a random screenshot from vice city. The upper floor of that pay 'n' spray building is more-or-less a single cube with a texture (I think it is just a photograph of a real building).
The video you sent me, he does a lot of sculpting in the previous videos in the playlist, and ends up with a model with 39 gajillion tris, which might be fine if you just want to render a picture with blender, but if you want to put it into a game without completely annihilating FPS, that's going to be too much.
>you dont need a separate material for each screw, a single material would probably suffice for all metal parts
Yes, sorry, I didn't mean material, but you still expect me to actually model the screws and place it everywhere I need it, right? Let's suppose I have a plastic stuff held in place by metallic screws. That's two materials, but that means I need mesh for the screws, because otherwise I can't assign the metallic material to anything. But then even if it's tiny and ingame you'll never be able to tell that it's not an actual screw just a texture on a flat plane, unless you move really really really close.
Replies: >>220602 >>220603
I played the demo and couldn't get past the first level I think.
You finished this?
And here I thought it was dead.
Just watch tutorials. 
Entire from 0 playlist and materials.
>The video you sent me, he does a lot of sculpting in the previous videos in the playlist, and ends up with a model with 39 gajillion tris, which might be fine if you just want to render a picture with blender, but if you want to put it into a game without completely annihilating FPS, that's going to be too much.
<You bake textures and retopologise the meshes. 
Just look it up.
Replies: >>220642 >>220687
Screws those are probably rivets but whatever should be physically present on a high poly model and baked into the textures for the low poly one. Materials on the high poly model are procedural, bake down into actual textures for the lower poly model which you actually use in the game.
>plastic with metal screws
Aside from that being an odd combination, especially in a game, you could use a metalicness map if it were important. If it's a minor detail on an unimportant prop no one's going to notice if the screws aren't shiny and you could just have the metalicness all as a constant based on the plastic.
>The video you sent me, he does a lot of sculpting in the previous videos in the playlist, and ends up with a model with 39 gajillion tris, which might be fine if you just want to render a picture with blender, but if you want to put it into a game without completely annihilating FPS, that's going to be too much.
Learn what retopology is my dude. Though according to Unreal, Soon™ you shouldn't need to give a fuck about polycount, they unironically want you to just throw megascans with a gorillion tris into your game and let Nanite take care of it.
Replies: >>220606 >>220642
I made it to the third stage, but I can't beat the boss, I spent over an hour playing it but I give up. However, I had fun and I respect what you are doing.
>Nice art style, particularly the character designs very reminiscent of like 2000s cartoons. I like the backgrounds too
>it controls pretty well overall, jumping feels responsive.
>the groundwork for something great is all there
<enemy placement can be kind of bullshit, like skeletons under keys that made me die a few times
<lack of checkpoints or at least recovery items makes it unnecessarily punishing
<HUD elements are a bit hard to read, the font used for the keys could be clearer and lighter since it blends with dark background or at least have an outline
<i need more audible feedback for when i take damage, also more i-frames
<talking to NPCs shatters the framerate until you die, no idea why this happens
Can't wait for games to come on 16TB RAM cartridges or be streaming exclusive
Replies: >>220607 >>220642
> or be streaming exclusive
It is what microsoft promised with their new xbox. But its likely just an excuse for always online drm, like it happened with some simcity "too complex to run on user pc, so it needs a connection to the server".
[Hide] (15.3KB, 500x500) Reverse
I suppose this is more me thinking that salty blizzard fans or execs that are pissed that OW2 is doing worse than TF2, but I'm thinking...

Corporations push boundaries for many reasons, many of them bad reasons too. Cheating and botting is technically not illegal, but mostly-everyone hates it for completely-justified reasons. On top of this, the right kind of bot service is cheap, and a failing company that has no moral boundaries or is desperate for recognition/profit, would absolutely use this to try and cut-out the competition.

So, have you guys ever found out instances of deliberate corporate sabotage of online games via bots? I was originally gonna post this in QTDDTOT but it seemed elaborate enough to warrant it's own thread.
Replies: >>220636

I can't believe i typed this up ONLY to have it posted in the AGDG thread.

Replies: >>220639
Thanks for the bump!
>muh retopology
I tried to look into it but from what I found it's mostly useful for "organic" materials, like rocks, animals, bread, etc. where you have a lot of "irrelevant" detail. Not sure about more blocky things, like what if I have a building or machine made from a few big metal planes, adding a bunch of screws or rivets will add a lot of useless geometry, and I don't think these retopology tools will make any good with them.
>according to Unreal, Soon™ you shouldn't need to give a fuck about polycount
Thanks, but I'm more interested in some early 2000s style, not google earth: the game.

>be streaming exclusive
That worked fantastically for onlive, stadia, and who knows who else tried it. Unless they manage to break the laws of physics, I don't see any of them succeeding for any game requiring faster reaction times than chess.
Replies: >>220644
>I tried to look into it but from what I found it's mostly useful for "organic" materials, like rocks, animals, bread, etc. where you have a lot of "irrelevant" detail. Not sure about more blocky things, like what if I have a building or machine made from a few big metal planes, adding a bunch of screws or rivets will add a lot of useless geometry, and I don't think these retopology tools will make any good with them.
I dont think you understand how things work. Unless you working in very specific way, you always need to retopologize. Its what you do, just like uv unwrapping, and texturing and modelling, and baking textures, its always a thing you must do. And rigging, if its something animated.
>screws or any useful geometry
Normal maps baking. https://youtu.be/mZqdDwyN2Xo
> I don't think these retopology tools will make any good with them
First of all, the usual "tool" in this case is you, you are the one who supposed to retopologise things. There are automated solutions, but they are not ideal, and never ideal for every situation. This is why companies, sometimes, dont optimize their shit, it takes time, so they just use insane models instead. And this is why UE pushes for it, pretending that it will work perfectly, without need for a company to spend money on it.
Replies: >>220646
>I dont think you understand how things work.
Maybe. But I don't think when they made games in late 90s/early 00s, they used all these procedural textures and bake texture shit (except to bake lightmaps). Yet they still managed to make a game. I'm pretty sure many times they just took a photo and slightly touched it up in an image editor. Or is there a secret node setup that will procedurally generate a vending machine for me?
Replies: >>220651 >>220661
Doom used 3d scans, digitized photos. After that they probably used generic noise textures, and applied them, which is equivalent to lowtech procedural noise, but worse. Gradient from white to black is a procedural texture. Checker texture is a procedural texture. Nowadays we just generate them, with infinite precision, instead of using an image file, because its faster anyway. And afterwards we bake it into image file anyway. This is how everyone does it today, because its better. Endlessly repeating bricks is a procedural texture. Filling the screen with the same color everywhere is technically a procedural texture. Using code to make a patterns, is what they are. 
Just watch the fucking videos. If you dont like how they do things today, just make your own 3d editor, this is how games were made back in a day. I assume, you think that back in a day, a small teams were able to make a lot of money, therefore old ways should be both more primitive and as consequence easier, less complicated. They are not. Modern technology makes it extremely easy, compared to the old ways, but you still have to do the fucking work. Retotologizing, UV unwrapping, weight painting (for animation), texture backing take a lot of (very boring) time. Just exporting a default cube, would take you like 5 minutes of work, of just pressing buttons, typing names for shit, etc.
Replies: >>220656 >>220661
I guess I'll have to settle with a text only game then. This shit is too complicated for my tiny brain.
Replies: >>220660
Just put in effort, videos are extremely simple, you just have to follow their directions, they explain everything excessively. Its not like you have to read a book, which explains everything in extremely obtuse way, and use ancient version of blender with left and right mouse buttons switched, because "its better to do things the opposite way of every single software in existence".
Replies: >>220683
[Hide] (328.2KB, 540x540) Reverse
I recall doom also painted their own textures

retopology is important in non-organic because reflections still exist, the way the light catches onto the material in a PBR workflow is still something to consider, a car or glass bottles must have good topology.
With something like stone walls, toothbrush, doors - it doesn't matter as much to be perfect with the topology.
The High Poly > Low Poly workflow is quite common in modern 3D games, modern being around the time normal mapping were being introduced.
Still you can just subdivide model your way through for most hard surface objects but as soon as it's rerflective or very organic looking, you'll have to adopt a similar 3D animator mindset
Replies: >>220663
Spoiler File
(226.6KB, 530x263) Reverse
Replies: >>220714
Retopology is almost always important, unless you baking 3d>2d sprites. You have to consider time spent on it, vs time gained by doing it. If you want to render some complex animation (even if its just spinning object), it might worth your time to make sure object is optimized, if rendering animation will take more than few hours. But if you have to waste 30 minutes to optimize something, and it will save you just 60 minutes of rendering, its not fucking worth it. (unless you have to, because you dont have enough ram) Even for a game, lets assume you have player model with 20k polygons, and you can reasonably cut it down to 5k. You might think "quarter polygon count = quad performance boost!" but in reality you will go from 60fps to 60fps (or 300 to 301 fps), because the difference was below 1%. So optimization would be complete waste of time, especially for a single dev indie game. 

Also "clean" topology is more important than minimization of number of vertexes, in virtually every case, but "clean" topology leads to minimization anyway, but its still important to understand priorities.
[Hide] (824.2KB, 1280x720, 00:25)
This is the guy who posted the sword demo from the last thread. I cannot for the life of me figure out how to make the sword go in the right direction when the players throws it. it's based on where a straight-line raycast is from the sword to the player's mouse is (minus the X coordinates).
This is what I have so far, and vid related is what happens. Red box is the actual coordinates where it's being launched from.
$StrengthMeter.set_cast_to(Vector2(0, get_local_mouse_position().y))
linear_velocity = to_local(Vector2($StrengthMeter.get_cast_to().x * 10, $StrengthMeter.get_cast_to().y * -10))
I can post more screenshots later, right now I just needed a video of it.
Replies: >>220674
seems completely wrong, why even bother with a raycast?
you're already rotating the rocket sprite, so just use that as the angle and then move with vec2(cos(angle), -sin(angle) * vec2(mag, mag)) as the velocity
Replies: >>220677
What is mag?
Replies: >>220683
[Hide] (105.3KB, 782x1322) Reverse
I'm feeling motivated, I'm going to start working on my game about a bunny girl that kills the government. 
I already have the character design, a general idea of how the game works, and experience with godot. If I don't post again within the next five years it'll most likely be because I couldn't make the sprites look good and gave up.
Replies: >>220790
>blender with left and right mouse buttons switched
Um, actually I still use that way, since blender was the first 3d software I tried, and I got used to it. Something like liking GIMP's UI, when everyone else on the world hates it. Anyway, some random, tangentially related questions.
Why does he do that weird thing with a texture and making 0 sized UV maps. It seems a bit weird to store 3 UV coordinates per triangle, when all you need is one RGB value per tri (or probably just one linear palette index). Or is it just that using attributes is too obscure/cumbersome?

And one more random question: what if I have a model (downloaded from a site, ripped from a game, whatever) that's almost what I want but not exactly. Does that mean that if I don't have access to the original procedural texture setup/texture bake setup I'm fucked?

Magnitude of the vector. Practically the speed of the shit (vec2(cos(angle), sin(angle)) will give you a unit vector).
Replies: >>220686 >>220710
>Why does he do that weird thing with a texture and making 0 sized UV maps. It seems a bit weird to store 3 UV coordinates per triangle, when all you need is one RGB value per tri (or probably just one linear palette index). Or is it just that using attributes is too obscure/cumbersome?
he's taking an artist approach and this is what he can do from his pipeline, if you were the programmer & artist, then yeah you would use color attributes, write your shader to not use texels and instead pack the color data with the vertices instead, although it might become awkward to pack that data (verts, normals, uv) vs. (verts, normals, color)

>Does that mean that if I don't have access to the original procedural texture setup/texture bake setup I'm fucked
if you don't have the source file, then yes
Replies: >>220710 >>220718
I have no idea what he's doing since he's using the latest version and I'm not.
[Hide] (347.9KB, 1920x1081) Reverse
>Why does he do that weird thing with a texture and making 0 sized UV maps
Its not 0 sized UV map. he uses regular UV map and extremely tiny palette-like texture file. He uses texture files as small as just 4 pixels. Because its (coordinates*0) you think the result is 0, but in reality it works differently. You cant have less than 3 coordinates per triangle, because this is how triangles work, and his Zero, is actually a float value, which points to a coordinate in texture file. In picrelated you can see 2 boxes (on the left) placed on UV map, one is smaller than the other (and you can imagine just a single pixel dot there as well), but both of them point to a single pixel, despite their apparent difference in size. In the end you still have 3 coordinates, and in his case, all 3 of these coordinates are in the same pixel, and it gives flat color for the whole polygon. 
His way is probably the fastest way to do it, and he gets everything he needs for game engine, texture, model, uv unwrap, and bones for animation.
What you proposing is vertex colors, and they are practically useless for anything other than weight paint for animation. There is no way to make them look good, since you need way more geometry, because 1 color require 1 vertex, but virtually every point of model would be a gradient color between 2 vertices. 
>It seems a bit weird to store 3 UV coordinates per triangle
Its the only way to do it, unless you want to make your own engine and pipeline. 
>Does that mean that if I don't have access to the original procedural texture setup/texture bake setup I'm fucked
Basically yes. There are ways to deal with it, but it is very cumbersome, and might take more effort than remaking whole thing from scratch.
>if you were the programmer & artist, then yeah you would use color attributes
No, you would use his approach, because writing a separate shader, and typing in colors and coordinates by hand, for each model is insane. His approach is the same approach everyone else uses, model+uv+texture file. Unless you rendering on a literal calculator, it is the fastest way to do it. It just looks strange, because of how tiny texture files are, but it is in fact regular way. 

You look at a guy modelling, texturing, unwrapping and rigging a dragon in 10 minutes and think to yourself "actually I can be more efficient, instead of wasting a single minute to completely color a model, he could invent some other technique which would take even less time".
Replies: >>220711 >>220718
>No, you would use his approach, because writing a separate shader, and typing in colors and coordinates by hand,
you could do that, but you could write an intermediate program or a blender addon to do all that.
as you export your model in whatever format, you just read the texel for each vert and then put it back as either a number for the palette idea or an rgb value and potentially roll your own fancy format or do it as you're loading the model in for some very autistic reason.

Regardless, I would still stick with the usual pbr/god shader as it's not worthwhile apart from stroking some autism want
Replies: >>220713 >>220718
[Hide] (330.2KB, 452x512) Reverse
Let me expand a bit.
What probably confusing you is scaling to zero. You assume the result of it is zero, but its not. You have 4 coordinates, for a quad. (same principle for triangle or any other polygon)
a(0.0 ; 0.0) b (1.0 ; 0.0) c (0.0 ; 1.0) d(1.0, 1.0) 
When you scaling them to 0, you dont erase them, and in this case you move them to the middle of every related coordinate, ie a=b=c=d= (0.5; 0.5). UV values are always stored like that, even if texture is 1 pixel wide. 
>you can write multiple programs to convert from model+uv+texture, to get model+uv+texture
Or you can use model+uv+texture you already have, and do things the same way as everyone else.
Replies: >>220718
[Hide] (219.2KB, 474x266) Reverse
[Hide] (697.4KB, 1922x1064) Reverse
>awkward to pack that data (verts, normals, uv) vs. (verts, normals, color)
SOA is always an option, and then you don't have to care about packing.
>What you proposing is vertex colors
Not exactly, I was proposing per face colors. You could do with per vertex colors, but then when you have neighboring faces with different colors, you'd need to duplicate the vertices in the same position but with different colors, but that just plain sucks.
>Basically yes. There are ways to deal with it, but it is very cumbersome, and might take more effort than remaking whole thing from scratch.
>because writing a separate shader, and typing in colors and coordinates by hand, for each model is insane.
You don't have to type in colors by hand, you can add arbitrary data to vertices/edges/faces. I don't even need a fancy custom game engine, per vertex colors in blender, it took like 5 minutes to set up, and 90% of that was figuring out how to use custom attributes.

>I would still stick with the usual pbr/god shader
Why, I don't want unity standard shader look, nor do I look for some hyper realistic shit. Old games did fine with gouraud shading, and you don't need the whole pbr kitchen-sink if you're looking for some cel-shaded look.

No, my problem is not that. My problem is storing 3 UV coordinates (6 floats) when you only really need one RGB color (3 bytes), or just a single palette index (1 byte).
Granted, today when even simple games need gorillion of gigabytes of VRAM, it probably doesn't matter that your 1 kbyte low-poly model uses a few more bytes of memory than necessary.
Replies: >>220722 >>220727
[Hide] (150.5KB, 346x732) Reverse
[Hide] (43.5KB, 400x150) Reverse
>it took like 5 minutes to set up
One system uses blender specific api, which is not present in any modern engine. And other is generic system used in every single engine, and exporting it to a game engine takes no additional work, its 1:1 game ready. Why do you try to invent the worst and most backwards ways to do things?  
I already posted a bicycle with square wheels.
>gouraud shading,
i recall it was blinn-phong was the lighting model back then
for the most part I think PBR is pretty good overall, but that doesn't mean you have to abide by it fully.
you can start from PBR, then kill AO and make lights throw their energy linear instead of non-linear, then you'll be closer to a more classic look on the lighting and then take all of the modern advances in shadows for example to take things further
[Hide] (11.8KB, 401x232) Reverse
I went to another game dev meetup last night where a friendly wizard who used to work with Godot solved my problem in one line:
linear_velocity = Vector2.UP.rotated(rotation) * ($StrengthMeter.get_cast_to().y * 10)
So that's nice.
Then after a while some stronk womyn of color showed up and were talking about making a game about something but I couldn't hear from the far end of the table, I kept hearing "administration" so I assumed they wanted to make a game about business administration or finance so I suggested something like SimCity as a reference point.
Then as the conversation went on I realized what they were actually talking about.
They weren't saying "administration", they were saying "menstruation". They wanted to make an educational game for young girls about periods.
>mfw unironic discussion of a RPG based around periods
Replies: >>220752
[Hide] (84KB, 471x464) Reverse
One wizard is worth enduring the chattering of a hundred females.  Good on you, anon.
[Hide] (600.9KB, 1024x600, 00:18)
Are there any fun 2D games with non-linear movement? 
My original idea was to have walking and running be sort of like tiny bunny hops that speed up and stop a few times per second, probably using abs(sin(time)) or something to that effect. Since the game is meant to be fast paced I'm worried that having unreliable movement could make the game harder in a bad way.
[Hide] (97.2KB, 613x480) Reverse
[Hide] (159.2KB, 1024x600, 00:05)
And another question, how do I design feet? 
I got the rest of the design to look how I want it to but I never really thought about rabbit feet before.
>how do I design feet
1) don't be a footfag
2) learn to into anatomy
3) decide if you want wedge-like Bugs Bunny feet, semi-realistic furry digitigrade feet or the traditional cop-out of shoes
Replies: >>220849 >>220852
[Hide] (245.1KB, 850x649) Reverse
[Hide] (672.6KB, 1920x1081) Reverse
[Hide] (675.6KB, 1920x1081) Reverse
That would completely depend on level design. But most every game already figured out that dash+ double jump is the way to do it. 
I would say hind legs is the way to do it. It can have kind of crouched pose normally. And it straightens during jump. Kind of like picrelated. Maybe a bit shorter torso and a bit more meaty legs. Or lower ass. It probably doesnt matter.
Replies: >>220849
Spoiler File
(597.8KB, 974x682) Reverse
>don't be a footfag
I think my problem is that I'm not a footfag. I have a clear image of what the legs should be like and now I'm just trying to come up with feet that will blend in and not be distractingly ugly. Nothing really sticks out from existent designs as the type that would work. 
Small feet on thick legs look weird but big feet in general look retarded. 
I don't want to admit defeat and rely on boots.
>That would completely depend on level design.
It's meant to be a fast paced combat game with little platforming. Like a beat em up but with good combat and fast movement. 
The hopping walk cycle was meant to be a fun gimmick but I'm worried it could end up getting the player killed too frequently.
>I would say hind legs is the way to do it.
That's sort of what I'm going for. Thick hind legs for big bun jump. I just can't figure out what should go on the ends of the legs.
Replies: >>220852
>reeeee footfags!!!!!!
completely ignore this retard

feet are really hard, they are like a perfect ratio balance and have a ton of shape, it's not like drawing an arm where you just draw a few lines and you're done. toes have a correct length from stubby to somewhat long, anywhere in that range should be fine but look at reference to make sure you're not incorrectly making the middle toe the longest or something. also pay attention to their shape. for soles you need to look at the shape of the ball of the foot, it is "compound curve" shaped and as you go down the sole you need to account for the arch, you do not want a flat foot. the foot becomes more narrow towards the heel.
viewing from the instep of the foot, it should be straight and make its way up the ankle a bit, you will need reference for that so that it looks right. on tip toes this location does not change, you simply point the angle down and make the toes go at a harder 90 degree angle, and the arch will be more exaggerated from this angle and the bottom of the heel should point outward a bit. the front of the foot can be tricky as well, on the arch side its mostly a straight drop down but on the side with the little toe it gradually curves out a bit. remember to get toe sizes right and not draw them like fingers.
use foot fetish art to your advantage, there are tons of good drawings of feet that you can find with tags like "foot focus" and this should help you find foot fetish artists who, due to having a foot fetish, are obviously going to be picky about drawing anatomically correct feet. you can use it as good reference.
Replies: >>220869 >>220873
[Hide] (262.8KB, 1200x1650) Reverse
[Hide] (24.2KB, 563x516) Reverse
[Hide] (87.6KB, 960x852) Reverse
[Hide] (37.5KB, 600x400) Reverse
[Hide] (104.4KB, 1080x1080) Reverse
here's reference to help you
[Hide] (270.5KB, 1260x766) Reverse
>look at footfag art 
I'm not sure if that's what you intended but searching for foot_focus on e621 gave me some good design ideas in exchange for part of my soul. 
For now I'm going with the ones on the right. It's a mix of some twink catboy's feet I found and a real tiger's paws, but could believably pass for the front paws of a rabbit. 
Video games.
Replies: >>220870
Spoiler File
(4.1MB, 3024x3144) Reverse
Spoiler File
(2.2MB, 3024x4032) Reverse
Spoiler File
(2.2MB, 3024x4032) Reverse
Further research material for your consideration because I won't soon again have an opportunity to post them.
Replies: >>220872 >>220914
Isn't it painful to stand with that pink plastic(?) things on your feet?
Is that you?
Replies: >>220875
Speak for yourself, it's established fact that footfags can't draw feet. Their brain damage prevents it.
It's foam, and no it's not me.
[Hide] (1.5MB, 720x540, 00:24)
[Hide] (4MB, 720x540, 00:37)
Suwako from touhou 12.3 don't laugh. Vid 1 shows the animation. Vid 2 shows the discrete nature of her movement.

Her forward and backward hop show two different approaches to aerial state and hoping:
- Forward hop puts her airborne so she can only do things she usually can in the air, e.g. aerial attacks, air-dash, or flight. Specifically, she can't jump until she lands.
- Backward hop keeps her in a grounded state and so can be cancelled into anything she can do while grounded except another hop. So things like grounded attacks and jumps all work. It's similar to other character's walks in terms of the players ability to interrupt it.

For speed, her movement is slow paced making it easy to know and plan around what state she'll be in. She could probably hop quite a bit faster and have it still be manageable, but you don't want a players jump input to come out 0.2 secs later due to a mechanic they are disinterested in. On the other hand, it there is another mechanic which encourages the player to pay attention to their hop state, I think it could work. Or maybe just having a nice rhythm or sound effect would be enough.

>rest of the thread
oh no
Replies: >>220914
This thread just won a 【LEGACY OF SOKUNIGGER】 award.
Replies: >>220926
[Hide] (233.6KB, 400x343) Reverse
Replies: >>220914 >>220923
[Hide] (118.7KB, 1358x518, 00:05)
[Hide] (103.7KB, 1358x518, 00:03)
That's a guy, isn't it?
In my case it wouldn't be that extreme. Walking and running would still keep you grounded and allow for all commands to work instantly. The only differences between a normal walking cycle being that be that movement wouldn't be linear, making the hitbox effectively sway bath and forth while moving, and the player would still keep moving until the end of the hop unless cancelled with another animation. 
This CGI recreation shows the sort of speed and height I'm thinking of. 
I want my bunny to be a cute death machine. The kind you want to pet but can easily imagine killing a dude. Nanachi is just a cute onahole with no violence in her.
Replies: >>220918 >>220938
>embedding raw youtube URLs that have nothing to do with anon's statements
What third world country are you from?  Kill yourself.
[Hide] (2.2MB, 2048x1152) Reverse
If you're worried about movement try playing games that do it well like the sonic CD games, and avoid stuff that felt somewhat awkward like Jazz jackrabbit. Although jazz was still fun to play to, I just found the movement a little disorienting compared to other platforming games. Bubsy is also a great example of how NOT to do it.
Nanachi sure is popular.
Replies: >>220938 >>220984
Only way to escape the prison is to make another one.
[Hide] (809.4KB, 1280x720) Reverse
>I want my bunny to be a cute death machine. The kind you want to pet but can easily imagine killing a dude.
Replies: >>220938
>The only differences between a normal walking cycle being that be that movement wouldn't be linear, making the hitbox effectively sway bath and forth while moving, and the player would still keep moving until the end of the hop unless cancelled with another animation.
The fighting game Skullgirls has a character called Parasoul who has a unique walk cycle with what you're describing.  She takes these long, delicate steps, then leans her entire body forward once her front foot is planted.  It's kind of like a waltz.  Walking with her is odd since holding the stick for X time will move her very little, but X time + a short amount of time will move her a lot.

This movement is never relevant due to the pace of Skullgirls's gameplay (which sucks), but it's the only comparison I can think of.  I don't think having this would add anything to a platformer, but it may be a fun way to reinvent the wheel and make your game stand out.

What the other anon described about the Touhou frog girl in Soku seems like it may lead to more intriguing gameplay, where your character can only move in discrete increments and you need to position yourself very carefully.
Replies: >>220938
[Hide] (4.5KB, 159x163) Reverse
I can't fucking decide if I want the hops or not. I think it looks fun by itself but I'm still worried that it could screw the player once enemies and fast combat are a factor. 
Pls "play" and gib opinion. Space to toggle hop mode and the rest is what you would expect. 
files.catbox.moe/wm88dv.7z (windows) files.catbox.moe/jzant6.7z (linux)
I'm not going to bother with animations for now, I just want to figure out the basic mechanics and try them out. 
I don't really know of any sidescrollers that have movement I enjoy. Jazz jackrabbit is the only game that comes close to what I want in fluidity and speed, but not in controllability. 
It's a casual bun, those are too flashy. 
It does sound similar. I'll give it a try and see how she plays. 
>I don't think having this would add anything to a platformer
That's sort of the intention. The hopping is there mostly for the cute animation, not to be a game changing mechanic.
[Hide] (4.7MB, 1280x720, 00:13)
[Hide] (3.9MB, 1280x720, 00:13)
>I'll give it a try and see how she plays.
Please do not actually play Skullgirls.

>not to be a game changing mechanic
>please play it
I downloaded and messed around with it for a few minutes.  The hop obviously feels better because it's faster but there's something unusual about the visual of the character going up and down, yet not actually being airborne.  You may be able to cover that with an animation like the MegaMan X games where X puts his entire body into a "run" but barely moves at a snail's pace.

Maybe the walking could be normal and the run could be a hopping function where the character moves much faster but is harder to control, and has to skid to a stop.
Replies: >>220965 >>221047
[Hide] (35.7KB, 500x363) Reverse
Made some small improvements to my PS4 emulator this week:
>Added instruction operands' types in the GCN disassembler
Thought about writing a script to generate data tables for this, which turned out to be a bad idea since the tables were big and unreadable and the script was bloated as well.
>Unified register loading and saving in shader translator
Which means less code used.
>Re-added hash and caching of constants and types in SPIRV assembler
It was a stupid idea getting rid of this in first place as without it translated shaders can get quite large.
>Fixed out of bounds access in shader analyzer
This caused a stack corruption and debugging it took way too long, even with Address Sanitizer.
If you find yourself in this situation, make sure your compiler isn't inlining any functions you suspect are at fault with attribute((noinline)) or similar, otherwise the stack sanitizer might not work properly.
I don't think the hop needs to be implemented in the code. That is mostly dictated by the animation.
I do think it would become jarring specially if that's the max speed.
Replies: >>221047
[Hide] (59.1KB, 638x532) Reverse
this reminds me of worms movement but not as quick

you could do
crouch, then forward (236 or 23)
the longer you crouch, the bigger your hop
the faster you flick the movement, the bigger your hop as well
whether or not you do a half circle, quarter circle, how big your circle for that matter
then you can include up to do a bigger hop or make it 'taller' instead of 'hoppier', like 8236 vs 23698 (up, crouch then forward vs crouch then forward and up)
it would be pretty kinetic, unique and a bit goofy

it would be so goofy and retarded that it could work
Replies: >>221047
>Bubsy is also a great example of how NOT to do it.
Explain this. I always thought Bubsy movement was just Sonic movement with the physics constants ass-backwards and the problem was more the level design making it painful to actually go as fast as the game lets you.
Replies: >>220991
Bubsy has basically 3 speeds: stationary, crawling, and mach 7.
I would like a traditional 6 button fighting game but with jump on its own button instead of Up on the directional dpad or stick.
Yes before you ask, I play grapplers.
Replies: >>221021 >>221047
[Hide] (333.8KB, 1080x1920) Reverse
On the off chance you aren't baiting, just get a leverless controller or use a keyboard.
>i play grapplers
What you mean to say is "I play exclusively Capcom games and I have internalized that it is normal for a game to have a 360 motion despite it being absolute garbage."
Replies: >>221037
[Hide] (32.3KB, 210x196) Reverse
[Hide] (500.1KB, 600x1047) Reverse
[Hide] (221.8KB, 330x400) Reverse
[Hide] (1.2MB, 1200x1166) Reverse
[Hide] (5KB, 182x207) Reverse
>abloobloo muh gabgom
Replies: >>221046
>traditional 6 button fighting game
[Hide] (57.8KB, 1024x600, 00:04)
[Hide] (543.1KB, 1544x960) Reverse
I got a question about gore this time. Are there any good ways to make permanent gore splatter in 2D? 
My first instinct would be to just draw gore all over the map and have it tied to an opacity mask that's "painted" by particles as they move, though I'm not sure if that would be faster than just having 200 sprites laying around. For context the game is going to be screen based with each screen having around 50 enemies at worst. 
Also, the cunny is clothed now. I kept going back and forth between making clothes too slutty and then making them cover too much skin until I eventually settled on the diamond dress. I'm aware she's leaning, I noticed 2 clothes in. 
>Please do not actually play Skullgirls.
I bought it at launch. I'm already damned.
>Maybe the walking could be normal and the run could be a hopping function
I think I'm going to have the hop be an alternate movement between walking and running. If you hold jump and a direction when landing the character goes into hopping mode instead of walking and it's slightly faster. Maybe with a fast hop if you land from a long jump. 
It's not that complex anymore, all it does is take the y position of the animation and multiply it by the speed. I think it's better that way as it keeps the hitbox centered around the sprite. 
I plan on aping fighting games for the combat but not for the movement. Complex inputs work well in enclosed spaces but I would kill myself if a platformer required me to 623 constantly. 
What's stopping you from remapping up?
Replies: >>221049 >>221079
[Hide] (907.5KB, 1263x568) Reverse
>I bought it at launch.
That's probably the most reasonable time to buy it, but I still pity you.
That sounds like a cool idea.  Kind of reminds me of the running in Pizza Tower.
Replies: >>221070
[Hide] (67.5KB, 309x421) Reverse
Another demo because I'm extremely indecisive. 
First one requires you to hold jump after landing to hop. Second one requires you to hold down instead. 
I think holding jump feels faster somehow, but the second one makes it easier to chain movement. specially high jumps. 
>Pizza Tower
That reminds me, I still need to play that game. I heard a lot of good things about it.
Replies: >>221078
>I still need to play that game
It's probably one of the best 2D platformers of the last decade.

Of the two demos you posted, the one where I hold jump to hop makes more sense and feels better.  There's no crouching state, and hopping is jumping even if your game logic says it's just a weird type of running, so that's what my brain feels is normal.  Holding down doesn't make sense, although that may be partially attributed to the game not having a crouch/crawl input yet.  If down did other stuff, then it may be more intuitive to assume it would modify movement, but that may not be what you want to do.

I don't understand what you mean by the hold-down-to-hop method making it easier to chain into other types of movement.  I can't go from holding down + direction to tapping up in order to do a high jump.  That may be due to the high jump down -> up inputs being a pretty small window but the way you phrased your conundrum makes it seem like there's something else you intend for the bunny to be able to do.

I like your drawings, by the way.  Keep it up.
Replies: >>221123
Your idea gave me an idea, not sure if it'd work for you but feel free to use it if so: really slow walking with just movement input, automatic pretty fast hopping with movement + a hop button held, and the fastest movement when you move and tap the hop button with the right timing. 
Then only require extra fast movement for some platforming sections and to make dodging stuff easier, thus requiring at least some skill, while speedtrannies will be tapping the entire time so hopefully they all get arthritis after 500 runs and fuck off.
Your idea sounds like what Super Meat Boy has, and it works well with simplistic blood, though if you want more high detailed stuff like the word gore implies to me at least I'm not sure it'd work very well. Combination of the splat layer and objects is always an option, though if you want it to be persistent between screen reloads that's a lot of object serialisation to keep saved.
Replies: >>221123
[Hide] (308.8KB, 1024x600, 00:12)
[Hide] (434.3KB, 1024x600, 00:13)
Took a lot of double-viewport spaghetti but it works now. It's "simple" enough that it shouldn't impact performance too badly, hopefully it'll look good with non-placeholder blood. 
>That may be due to the high jump down -> up inputs being a pretty small window
You can hold down while pressing up. Maybe the retarded way I place my hand on the keyboard is why I think it's easier. 
Either way I agree that holding up feels more intuitive, I'm going to keep it that way for now. 
I think that would make movement too complex for combat. Platforming is meant to be more incidental with the focus being on just killing things and turning the screen red while not getting hit. 
Though it's ironic that the demo gave you that idea, that's how dash cancel works in EDF 4.1 and that's one of the inspirations for the bun game. 
>Your idea sounds like what Super Meat Boy has
It seems to be. Hopefully that means it's not too expensive to run. 
>Combination of the splat layer and objects is always an option, though if you want it to be persistent between screen reloads that's a lot of object serialisation to keep saved.
I might go for that if the splats don't look good enough by themselves. If I let the player backtrack I'll probably just save the splat alpha as an image and despawn the objects.
Replies: >>221125 >>221231
>Maybe the retarded way I place my hand on the keyboard is why I think it's easier.
I sincerely hope you aren't one of those "thumb on down" people.
Replies: >>221127
>thumb on down
Hell no. 
My hands move around a lot but usually index finger on left, middle on down, and ring on up when going left, index on down, middle on up, ring on right when going right, or index on left, middle on up/down, and ring on right when alternating.
Replies: >>221240 >>221242
Trying the donut tutorial again, wish me luck. If you don't hear from me in 3 days, I probably got lost in blender's labyrinthine UI and starved to death, or I got hungry, ate the LCD monitor and died of poisoning.
Replies: >>221277
[Hide] (17.6KB, 542x655) Reverse
I hope to see where this goes, bud! Love me some gorey bunny
Replies: >>221540
[Hide] (603.6KB, 614x498) Reverse
Absolutely disgusting.
Replies: >>221540
arrow key master race
Replies: >>221540
[Hide] (1.6MB, 1920x1080) Reverse
Dunno, maybe? I'm not going to wait for hours to render a video
Replies: >>221280 >>221306
You dont have to. Its just "proof of concept", you only need a single render to know you can do it. And you can render in low resolution/eevee or even workbench. Also, dont bother rendering into webm/video formats. Its bad idea, and way worse than a lot of png images, you can stich them later(or never).   
And you most certainly dont need a lot of features. I dont remember if modern blender has "samples" option but it should be relatively low. I dont thing you need more than 30 seconds to render each frame, with denoiser. And you dont need subsurface scattering either. And you only need a couple of bounces. 
>bad ui
Its actually very intuitive, F3 menu brings almost everything you ever need, and everything else has search bars. You just need to know what you need / can do.
Replies: >>221295 >>221345
That definitely looks weird, especially that compositing works while everything else is the default white blender material. But AFAIK it's not really meant to be used for rendering anyway.
My main problem was anyway that the tutorial guy rendered an image in like 5 seconds, while it took me about 1.5 min with the same settings. I guess AMD still can't make a working GPU/driver, or that dual Titan RTX is ridiculously overpowered.
But to be honest I'm not too interested in doing animations right now, creating static models/environments are probably still enough challenge for me.
This one looks comfy as fuck, but I guess I'll have to find something way simpler: https://yewtu.be/watch?v=nb6rSMAooDs
>bad ui
I never said that. Actually, for the complexity and number of options blender packs, I think it has a great UI, but still, if you're unprepared you can easily get lost.
[Hide] (1.2MB, 1920x1080) Reverse
>forgot to attach image
Looks significantly better than other Blender doughnuts I've seen.
Replies: >>221338
But I just followed the tutorial
are you sure you were rendering it at the same resolution as the tutorial guy? like >>221280 said you can just render to a 360p window to get the idea of what it will look like
Replies: >>221347
He used some kind of nigger tiktok format of 1080x1440, while I stayed with 1920x1080. It's a bit bigger, but not that much.
But anyway, I don't think I'm going use cycles much, so it's probably not a big problem.
Replies: >>221356
1080p still takes a while, 480p or even 360p should give good results without taking too long
[Hide] (215.8KB, 604x548) Reverse
Working in blender is main source of income for that guy. And he has optimized settings. BTW you rendering with gpu? You might need to turn it on, and newer blender versions dont support any old radeon gpu. If you have old radeon, only older 2.93 version of blender still supports it. 
It actually has a lot of options, and while its not made for rendering, it can be quite useful.
Replies: >>221415
[Hide] (318.2KB, 814x1040) Reverse
>rendering with gpu
Yeah, I can see in radeontop when it's rendering, and actually halved the rendering time the first time I tried it (but it was with a much simpler scene). It can't seem to reach 100% though, usually it's around 80-90% usage which is a bit weird. Also after blender restart, sometimes I have to disable and enable GPU rendering, because otherwise it'll just use the CPU while displaying GPU in the settings. Oh, and this is after I manually compiled the hip kernel, because hipcc command line changed and it just throws an error, but at least now it does something instead of just crashing the whole blender like with the previous rocm version. TL;DR AMD still can't make a working driver.

Also, I hate chairs.
Replies: >>221496
[Hide] (1.3MB, 1920x1080) Reverse
Fuck this guy.
The texture he links is no longer available for free, Whatever, try to search for some poliigon dumps on torrents and whatever, don't find that image. I was also thinking about screenshotting the texture out of the video, but he only really opened the base color texture in somewhat passable resolution so give up on that too.
Find the texture on sketchfab: https://sketchfab.com/3d-models/wood-fine-veneer-oak-large-002-89cdeadb9e3548689031e9bdd8538990
No download button, but there's a webgl demo, it has to be somewhere in the network tab of developer tools. Actually no, they send you a texture with positions mixed up, as a kind of "encryption". Erm, whatever, just try to run chromium under renderdoc. Actually a bad idea, because renderdoc is a piece of shit and doesn't support half of the opengl and generally shits up itself on anything but the most simple demo application. Try also apitrace, it successfully creates a trace, but crashes when trying to play it back. Fantastic, try the same with palemoon, the sketcfab preview fails inside apitrace (but wonky even without that). Anyway, chage plans, I have a qemu vm with virgl, let's run that shit under apitrace (I also tried renderdoc, don't know why, I wasn't even surprised when it failed), open the sketchfab in chromium in the vm, then spend the next half hours trying to figure out where the texture I need is in the 250MB trace file. It would be useful if this fucking apitrace would have a dump textures command, I'll have to look into that someday. Anyway, after all this shit, I finally got the texture (only base color, normal and glossyness textures though), with a watermark. So I got some extra work with UV layouts to make sure I don't use the parts of the texture with the watermark. If this goes on, I'll have to figure out how to patch chromium so it dumps all the textures (I guess I'd just have to somehow find the code for webgl's glTexImage2D wrapper.) Aargh, why can't things be simple.
Anyway, now I only have to figure out how to make these screws not gay, then I'll have a model of a fucking 650EUR chair, which actually has a downloadable .obj file on their webpage.
Replies: >>221500
[Hide] (1.5MB, 1920x1080) Reverse
[Hide] (75.4KB, 800x497) Reverse
[Hide] (204.9KB, 3000x1537) Reverse
Wood is very easy to make. Mapping +Noise+ramp with proper colors. 
Here a way to steal textures from any photo. https://youtu.be/gz4qV5_zuMo
>fuck this guy
Yeah, a lot of guides are actually commercials for their asset packs. Man gotta eat, but they really push it with their "1 minute to make X, just download X from our website!"
Here some cc0 stuff. 

If you learning blender, and specifically modelling, I would advice modelling cars. They have a lot of detail and perfect 3d projection images, for references. I use a little trick, bevel with very low value, like 0.01cm, and subdivision surface on top of it. It makes perfectly smooth and sharp edges. 

Look fine to me. Some of them might be dark, due to not being illuminated by any light sources, just use hdri.
Replies: >>221521
>procedural textures
Yeah, but I was trying to follow a tutorial, and I didn't feel like doing a 2 day detour and researching procedural textures. And while the one you linked looks fine from a distance, go close and it's a bit meh.
>I would advice modelling cars
But aren't they like fucking complicated? Thousands of details, with my speed and knowledge it would take months to model one.
The colors are a bit off, but my bigger problem is that right now they protrude out of the surface, when they should be in level. I guess I should get them on the texture somehow, otherwise I'll have to figure out how to subtract the screws from the wood (so they won't z-fight) without pulling my hair out.
Dunno, could help, I just went with the tutorial, but there are a few too hard shadow anyway... but right now I'm trying to learn to model, and maybe texturing, rendering will be a different beast (especially if you intend to import it into a game engine).
Replies: >>221568
[Hide] (3.1MB, 1024x600, 00:15)
[Hide] (42KB, 1636x854) Reverse
[Hide] (3.9KB, 190x125) Reverse
[Hide] (748.6KB, 00:56)
[Hide] (1.4MB, 01:19)
I organized the game files and did some autism.
Folder structure went from baby's first demo to a usable game framework. The way the overall logic works is more modular and less of a spaghetti tower. Level design is now completely separated from the gore system thanks to autistic scripts. 
>bun ost 
Bun got two awfully mixed songs. I composed the tracks and then recorded them with the most schizophrenia-inducing ghetto setup. 
Guitar goes to audio interface, audio interface goes to pod farm, pod farm goes to audacity. Simple enough, right? Except I need to use OBS to hear the guitar as anything else, including window's own built in audio system, is extremely delayed. Listening though OBS is the only way to get no noticeable delay, but only temporarily because there's still some delay and it's slowly getting slightly and slightly worse until you're playing like shit and the only way to fix it is to restart OBS. I'm also using Tuxguitar as a metronome, which sometimes stutters on beats for no discernible reason.
So I have this setup which is a ticking time bomb with my playing getting more and more delayed and a metronome that sometimes isn't on time. I made demo tracks for now but I'm going to have to fix this madness if I want to make real music. 
On a positive note audacity is pretty okay once you get into it. It's like GIMP before the graphical change, a really powerful FLOSS tool with a bad UI that's trying its best. 
I'm rather enjoying this autistic pseudo-layer system I'm using with paint. It works surprisingly well and I don't think I could easily replicate this style with CSP. 
My current plan is to make really smooth paint animations, like a 24fps walking cycle made in paint. The enemies don't have to be as detailed, but quality is imperative for the bun. 

Anyway here's the current demo. Next up is figuring out the combat system and then tweaking the movement. I probably won't post for a while until I figure that out. 
files.catbox.moe/kvhc3j.zip (win) files.catbox.moe/hykn3v.zip (linux)
Thanks. I'm hoping to get at least a proof of concept done with a couple areas and good gameplay. 
I already have a bunch of shit written down for bosses and the ending but my drawing skill still needs to catch up. 
Hey, as long as it works. 
Any two-handed keyboard only game should be played with the arrow keys. Or numpad.
[Hide] (66.1KB, 345x275) Reverse
[Hide] (2.3KB, 131x143) Reverse
>Psuedo Layering in Paint
I mean, whatever works for you buddy, but that seems really counter productive compared to using multiple layers in CSP to very easily onion-skin to compare and contrast, and avoid having to recolor and then clean up the reference (which I am assuming you're doing. 
You can still get that "binary brush" look in CSP too.
At the end of the day though, this is your journey, and part of the journey is the fun you have along the way, enjoying the process.
I can jive with this, guitarman. I actually find people using instruments cool as fuck
Suggestion: Make the blood particles way faster to match the visceral splattitude of the resultant blood splatters. Could even go as far to make the blood splatters instant through line traces to really communicate that violent momentum
Replies: >>221761
My current plan is to make really smooth paint animations, like a 24fps walking cycle made in paint.
Maybe you would like GrafX2. Has layers and can play animation to check how it's coming along as you work.
Replies: >>221761
[Hide] (463.8KB, 405x720) Reverse
>those sick homemade mixtape tracks
master, I kneel, teach me your ways
Replies: >>221761
Sounds like Hylics music.
Honestly, If you added drums to test.ogg it would be sick.
Replies: >>221567 >>221761
[Hide] (368KB, 800x600, 00:02)
Oh, quick tip
though to save you a bit of work/sanity.
Don't get too mega-tismal about like, drawing each and every frame of an animation to get it megasmooth.
You can get a significant amount of mileage just from shifting the asset's x/y position alone and abusing timing/holds. Plus, the smaller your sprite, the more blanks the mind will fill in and all that.
Strongly agreeable
Replies: >>221761
[Hide] (1.3MB, 1920x1080) Reverse
So I think I managed to bake the screw into the texture, the seat shows it, while the back side thingy still have the separate screw meshes.
Anyway, I hate things designed for artists, what is the way to create constants shared between different materials? You could create a driver, but that can only really be driven by object locations and I don't really want to edit booleans in a fucking 3d editor. Alternatively create a node group that only has constant outputs, you can share them between nodes. Argh.
And no if node. Mix comes close if you only ever call it with 0 or 1, but then you have true/false reversed. There is a script node, but that's cycles only and cpu only, so I wanted to avoid that.
Replies: >>221570
There are plenty of ways to share values, most everything is procedural, so same inputs will produce same outputs, no matter what. You can use image to store everything. And why do you need to share data between multiple materials? You can create a variable, and value-driver in textures, but usually you dont need it anyway. 
And no one will notice that your screws are one sided, no one will crawl around with magnifying glass to see if something is not 100% realistic and proper. Same with "if you look closely, texture looks wrong", it doesnt look wrong, it looks good enough for any situation. In many cases flat color works just fine, for example when you rendering forest, some people just use flat rectangles with somewhat random colors to represent leaves, no transparency, no sss, no textures, bump maps, just flat color, because it renders faster. 
>cars are complicated
Only on the first glance. They have schematics, so you just follow them. They have simple topology, no deformations, and they are basically boxes in terms of modelling. They are basically "follow the dots" images, you dont need any advanced techniques to make them. 
>The colors are a bit off, but my bigger problem is that right now they protrude out of the surface
So what, they look like screws and its good enough. You are not designing a perfectly comfortable chair. In fact, adding imperfections is a big part of modelling/texturing. but if you really want, you can switch coordinate system to "normal", and push screws into the wood
It takes like a minute to setup, and most of it is just picking hdri you like.
Replies: >>221572
[Hide] (210.6KB, 620x540) Reverse
[Hide] (54.7KB, 980x551) Reverse
>why do you need to share data between multiple materials
I was trying to do texture baking, but apparently it didn't occur to anyone at blender that people might want to bake position-dependent input values of the principled bsdf, so the way to do it is to disconnect/reconnect/change values with every material you use for each bake, which is good if you like gookclickers, but I'd rather have one constant to change and automatically apply it to every material instead of spending hours clicking randomly in the node editor. But to be honest, if I want to use baking for real, I'll have to write a little script that does all these settings mumbo-jumbo, bake the textures, restore the settings, and saves everything, because the procedure shown in every tutorial is fucking nuts. The computer should work for you, not the other way.
>Only on the first glance
I don't know. I tried to model an ac adapter lying on my desk (almost like picrel, except not usb-c), and I'm already not sure what to do. There's that cube, sure, there's that plug-shaped thing sticking out, kinda alright, but the topology is already starting to look weird... then cometh the two cylinder thing. I can't extrude a cylinder from there, so I guess I should add a new object? Unless I just create a cube and abuse the subsurface modifier. But maybe it would be better to separate this thing into 3 different objects. But then I'll have problems with the bevels between the objects. Aargh, I can't into modelling.
use a water tray, it stops the shit from drying out when baked
[Hide] (476KB, 1920x1081) Reverse
While backing is annoying in blender, the way you describe it feels very wrong. First of all, you use node wrangler addon? Next, you dont need to setup a lot of materials, you just copy your mesh(or retopology of it) and bake into it, into a single material. And I dont know why you need to randomly change values on materials, and even if you did, it takes very little time, since you can collapse node groups, and turn them into a single node, with inputs/outputs. 
>usb charger
The thing about them is the fact that they are very small and people will accept anything looking like usb adapter to be one. It looks fine to me, but small hole and prongs might be a bit difficult to make, but any version would do. Just use "i" to insert faces and adjust them (might need to press B and I a couple of times to get the insertion method you want). Scale them to size and use "circle" from loop tools addon, to make them circular, and than extrude them. Something so tiny just need to have vague shape of object you need, and it already looks almost perfect as is. If you want it to be smoother, use bevel object modifier and adjust it, but you really dont have to.
Replies: >>221594
>you use node wrangler addon?
>you dont need to setup a lot of materials
I might be doing something wrong, but in my simple case I already have 3 materials (one wood and two for the screws. And if I were to do that plywood stuff correctly, that would mean a 4th one. I'm baking it into a single one, but the source materials need to be set up correctly. And blender can't bake metallic correctly, so you have to set them to zero before baking (on each material), then restore it. You also can't bake a metallic map, workaround mentioned in jewtube is to put the metallic value into an emission shader and bake emit... but this means you have to fuck up your node setup on each of your materials, then restore. (I guess using diffuse shader and baking diffuse would work also). So I made a node group with a bunch of IsBaking* outputs, so I only have to change them, not each material in a pseudo-adhoc way.
>very small and people will accept anything looking like usb adapter to be one
Normally yes (and to be honest, if it's plugged in, a simple cube is probably sufficient), but let's assume I'm making a room escape game where you can pick up an usb charger and inspect it from close. I want to make a few models properly before becoming a lazy ass.
>loop tools addon
How many fucking addons do I need to make this thing usable?
>bevel modifier
Can I somehow adjust it per-edge? Not all edges need the same amount of beveling.
Replies: >>221598
First of all I want to say, that most of what you are doing is wasted effort and usually you dont need to do it, but in case of learning its okay. When I say "waste of time" assume "waste of time unless you learning".
>workaround mentioned in jewtube is to put the metallic value into an emission shader and bake emit
Its extremely simple, just a few clicks, ctrl+shift on whatever node goes into metallic, it will make emit node, bake it into random (1 color) texture, and than ctrl+shift on principled node to return it back. But having metallic map for such a tiny blips is a waste of time. Just reduce their roughness and they would look good enough. 
>2 materials for screws
A bit too much but whatever. 
> I want to make a few models properly before becoming a lazy ass
Time is finite, so you should know when not to waste it. No one going to give you a falling grade because tiny cylinders are actually hexagons. The only situation in which you have such small details is when you making some artwork, and it is a centerpiece. Or when you making animations for a product, so you want to make it appealing. And there are many situations where you dont want too much detail on mundane objects, you want player to ignore them, and focus on good stuff. 
>How many fucking addons do I need to make this thing usable?
2-3 dozens for casual use, and a few more for specific professional tasks. Blender has pretty much anything you need for 3d modelling, rendering, animations, etc. I believe there are free addons which generate whole medieval castle from a curve.   
>Can I somehow adjust it per-edge? Not all edges need the same amount of beveling.
Yeah, there are ways to do it, or just bevel it by hand, I believe its ctrl+B + mouse wheel, when edge select mode is active. There are also vertex groups/weights and creases and a million other things.
Replies: >>221599 >>221618
PS. also you can specify angle for bevel.
how do i stop being depressed and get motivated to maek gaem
[Hide] (223.4KB, 774x1030) Reverse
There's one faggot here actually making a game here >>221540 and some other fags having full-on Blender discussion ITT.
You're coming off as a nuisance with your petulant crybaby attitude. Stop making the thread about yourself. Faggot.
>just a few clicks
Yeah, but I still have to do it for every material. Maybe I'm just being spoiled with my developer self, where there is a script or bash oneliner for every repetitive and easily automatable task.
>Time is finite,
Yeah, but unfortunately my perfectionism doesn't allow it.
There's an indie developer I somewhat follow, he spew out a new 3d every 1-2 years, then I open the game, get to the second map, look at the wall, and there's a small hole in the geometry between the wall and the floor. Like wtf, that wouldn't pass even as techdemo for me... but that's probably why it takes forever for me to finish anything.
Replies: >>221620
make game while depressed and not motivated, motivation and undepression will follow
>Yeah, but I still have to do it for every material.
There are addons for blender to bake stuff more efficiently, and there are scripting support for every action you do, but baking is still the worst. You can "sidestep" it a little by making stuff easier to bake afterwards, but its huge pain anyway. I dont know about newer blender versions, but other versions required fucking "lights cage" setup (and no option for hdri), were dependent on rendering settings and often produced completely awful results. And never use "internal" texture, without creating a file, its a coin toss whenever it will be saved or erased from .blend file, and saving "internal" texture afterwards is painful as well. 
I remember looking for a better way to do it, but most other ways are just as shitty. Its a lot of manual work, which requires a lot of waiting for results, and repeating same steps over and over. "Substance painter" is a thing, but its non-free and provides same tools, and it feels wrong. 
Start with something easy and move to harder tasks. Starting from something daunting as making your own 3d engine is setting course for failure.
> And never use "internal" texture, without creating a file, its a coin toss whenever it will be saved or erased from .blend file, and saving "internal" texture afterwards is painful as well. 
not a complete coin toss, but you're better off saving the texture instead of remembering to do the "fake user" shenanigan
[Hide] (649.4KB, 470x843) Reverse
Are we back?
So I checked out this tutorial: https://yewtu.be/watch?v=0TzjpR3pdpk
One thing this guy does is that he selects a random face, Shift-D, RMB, P to make a duplicate into a separate object and work from there. But this way he ends up with a bunch of (partly) duplicate faces, and he only deletes a few of them. Isn't that a bad idea?
(Also the first time I made the building it was too low, normally I'd just go to edit mode, Alt-Z, box select the upper part, G Z and make it taller, but since this model was composed of 8 or so objects I couldn't do that (I don't think I can be in edit mode for more than one object at a time), so in the end I just scaled the whole shit along the z axis.)

Also, in the meantime I modeled a door. Baby steps. And I've managed to produce the worst looking procedural wood texture, I'll try to do something about that tomorrow. But my real question is, I want that door to look old and not exactly in a good shape. It should have paint, but that came off at various places (that's why I need the wood texture) (something like this, but with wood instead of metal, and probably without texture painting: https://yewtu.be/watch?v=bR4HtqxA9tk). I'll play around with my materials, but I think I should also need to add some deformation to the mesh itself, right? It looks too regular now, tree can expand or shrink as it gets moisture, etc. Or would it be enough to just add some random value to the normal's height input? And I'd need multiple instances of this door in a scene, but it would be better if they weren't exactly the same. Is there some easy way to do this, other than duplicating the materials and giving each door exactly the same setup but with a different seed value? (Also killing object linking in the process)
Replies: >>221695 >>221741
[Hide] (261.4KB, 1462x1080) Reverse
[Hide] (728.3KB, 3840x2160) Reverse
[Hide] (2.4MB, 3500x2333) Reverse
[Hide] (343.1KB, 1200x800) Reverse
For some reason I can't upload images.
I strongly recommend you download a courses for texturing in blender, it will help you familiarize with procedural texturing, the second link is a master class for a procedural brick.

As for the wooden door, personally I would start by gathering references, it will make your life easier, you need two types of references, one is a real world reference, for the door and the for the materials that you want to recreate. and the second type is final look of your product, go on artstation or other sites and search for wooden doors, these renders will guide you to final look you want.
>It should have paint, but that came off at various places 
You can combine the materials with a mix node
>something like this, but with wood instead of metal, and probably without texture painting
you can use large noise maps to place the wear and tear
>there some easy way to do this, other than duplicating the materials and giving each door exactly the same setup but with a different seed value?
You can use the random output from the object info node as value for the noise, you don't need to duplicate the material, at least is that what I would do.

More Resources. 
Replies: >>221713 >>222038
[Hide] (311KB, 413x788) Reverse
>all these resources
Thanks, I guess I'll be back in a month or two...
>gather resources
Yeah, should have done that. Like your third picture, it has two layers of paint (I guess it's safe to assume that old wood materials were repainted a few times). Also that yellow thingy, is that some fungus?
Picrel is what I came up on my own, but I guess it's more like an elephant cummed all over the door.
>an elephant cummed all over the door
10/10 horror
Replies: >>221732
Obviously it's too camouflage the door so zombies and other spooks can't find it easily.
Replies: >>221744
[Hide] (200.4KB, 410x389) Reverse
I must mention a software called Material Maker, its a MIT open source alternative for Substance Designer, personally I think the nodes are more intuitive than the blender nodes.
There is also an asset database with examples, so if you want to study complex materials its all there. 

I created this material in april using only this software, no external files,
Replies: >>221749
[Hide] (35.9KB, 549x329) Reverse
>,,, but it would be better if they weren't exactly the same. Is there some easy way to do this, other than duplicating the materials and giving each door exactly the same setup but with a different seed value? (Also killing object linking in the process)
see pic-related, you would use the uv mapping for the 'seed' of the procedural texture
for example, create seams for just the door frame (make sure it opens up like uv'ing a box), uv unwrap, then grab the door frame island and mess around with that uv island with scaling/translation/rotation
Replies: >>221749
Name something scarier than an elephant wandering the halls ejaculating on your furniture, your carpet, your vidya and your pizza.
Replies: >>221748
Furry conventions are terrifying I agree, but not in the way survival horror is meant to be.
Hmm, interesting, I'll note it, even though I'll probably stay with blender for the time being.
But that custom glsl node supports look interesting, that I have to do everything with these stupid nodes in blender can be frustrating at times.
Yeah, that's what I don't have now, since my door is full of mirror modifiers. Is it possible to somehow apply different UV coordinates to the mirrored parts without applying the mirror modifier (and losing the ability to easily edit the mesh afterwards)?
Replies: >>221752
[Hide] (32.2KB, 1920x1080) Reverse
[Hide] (18.2KB, 503x472) Reverse
Been working on stuff related to my PS4 emulator as of late:
>Writing a SPIRV virtual machine (for CPUs)
This would be useful to test PSB translated shaders without setting up any emulator bits (and perhaps the PSB shader compiler if you follow the crazy GLSL->PSB->SPIRV path).
So far it has a disassembler with a instruction formatting.
It could be too much effort for a problem of this size, so I'll probably develop it intermittently.
>Added custom primitive types sample to freegnm's examples
It just renders to screen all 5 unique (compared to Vulkan's) primitive types provided by the PS4's GPU.
There's another example program for indirect draws in the works.
In the end these should serve well in assisting emulation.
>Example programs now get their screen resolution dynamically
This was a patch submitted by an OpenOrbis developer, which was surprising.
He invited me to their Discord, but I can't/won't join it. Currently trying to figure another way to discuss with him and the team.
Replies: >>221757
[Hide] (10.4KB, 325x74) Reverse
If I recall, no
keep in mind you can look at the top right and click on the X/Y/Z symmetry during edit mode
When will I be able to play deniggered gravity rush on PC?
Spoiler File
(898.7KB, 1041x1079) Reverse
[Hide] (1.3MB, 864x1248) Reverse
>not loving the delicious brown
She's so fucking cute, and she has blonde hair and red eyes. RED EYES!
[Hide] (793KB, 1024x600, 00:07)
[Hide] (214.9KB, 1656x1811) Reverse
[Hide] (3.2MB, 1890x916, 01:01)
How simple can you make fighting game combat and still have it be engaging?
Currently I have 12 attacks spread over 3 buttons. I'm worried that adding more might make it too convoluted but I also don't want to make yet another brawler with simplistic punch/kick combat. 
That looks pretty neat. Paint is surprisingly easy to use once you get into it and abuse background colors and transparent selection, but I might go with that instead. 
>Make the blood particles way faster to match the visceral splattitude of the resultant blood splatters. Could even go as far to make the blood splatters instant through line traces to really communicate that violent momentum
That's sort of the plan, to eventually have the blood quantity and speed change with the attacks with the most violent ones just fucking exploding away. Blood lines are a pretty good idea. 
Keep learning songs until you start playing around with the instrument, and when you do play around make sure to pay attention to what you're playing. Being unable to recreate what you've just played is the worst mistake a beginner musician can make. That's what I've heard in a youtube video. 
When I'm trying to make a song I just go around playing single frets looking for the notes that sound right for the context and then go from there. Learning the shapes you like and knowing that scales exist also helps. 
I'm working on it. 
LMMS with Power DrumKit sounds surprisingly okay. 
I know about the basics of animation and I don't plan on rotoscoping the bun, but I also think making the animation as smooth realistically possible would be funny. 
[Hide] (1.3MB, 2480x2480) Reverse
Never because Kat is about as much of a nigger as Teagan, ie not at all because she's an ethnicity that doesn't exist in real life.
>How simple can you make fighting game combat and still have it be engaging?
See: Smash Bros
Freedom of choice of moves, as well as ability to go from move to move gives off that good vibe discovering optimal strings and shit
I like what you have going on for the mockup though.
Replies: >>221840 >>221917
I fucking hate AMD and their drivers.
>update system including rocm
>start blender, as soon as gpu renderer is enabled, it freezes
>kill it
>kill -9
<still nope
<dmesg: amdgpu driver tried to dereference a NULL pointer and it shat itself
>great. downgrade rocm to the previous version
>now blender sigsegvs. at least it doesn't fuck up the kernel, right?
>mess around for hours trying to figure out what else changed
>in a chroot with the old system snapshot with the same version of rocm it works, while I updated everything else
>double wtf
>copy the compiled binary to my host
>it works now
>I rebuild rocm
>it still works
So this piece of shit is so fragile that if you compile it multiple times, sometimes you end up with a version that compiles fine, but crashes instantly when you try to use it. Really, streetshitter nigger pajeets write better code than this.
I guess I'll really have to put this whole rocm and everything that uses it into a docker, because you only get a working configuration with rocm once every 5 years, and if you edit a comment in a config file it fucking crashes.
And before you say use novidia, yeah, it has cuda and that works, but that's all. If you want that, desolder the hdmi connector from your gpu so you don't even think about attaching a display to that because in the binary blob driver nothing works other than the most simplistic use case of using a single card with a single X server and not trying to any GPU acceleration technique invented after 1995. And intel is a fucking joke. WW3 can't come fast enough, I hope the first places that will be nuked on earth will be the nvidia and AMD HQs. Intel also, but for different reasons

On the other hand, I didn't have time to actually do anything with blender. Sigh.
Replies: >>221798
[Hide] (466.5KB, 431x125) Reverse
either you use shitty mesa drivers with intel chips that can't render anything more complex than a cube, nyidia cards that break xserver when you install their drivers (or enjoy having nothing work to nouveau), or amd which is held together with taiwanese duct tape.
i can't even build the latest AMDGPU kernel module for the latest Linux 6 kernel release, so I'm still using the one from 5. every time i apt-get it builds and fails.
pic related remains true forever
Replies: >>221804
[Hide] (38.1KB, 789x170) Reverse
>amd which is held together with taiwanese duct tape
works on my machine
[Hide] (350.1KB, 1024x600, 00:10)
[Hide] (381.4KB, 3314x1820) Reverse
I watched a few smash bros combat videos and I think what I was missing was high risk attacks with more damage. 
With a drop kick and charged attacks the moveset feels somewhat complete, now I just need to implement the attacks and work on some test enemies.
Replies: >>221874 >>221917
Keep the move count small for your own sake but worthwhile. Also figure out what type of attacks you want.
Generally, make sure moves always have a purpose, but isn't purely single-purposed but not over-bearing to become the go-to move for every situation.

Smash tends to have a lot of moves that stick out like a sore thumb for a long time, see jigglypuff's nair/bair or some of the neutral jump attacks in super turbo
where as mostly in fighting games, the attack comes out quick, isn't active for too long, etc.

I would recommend to focus less on COMBOS and focus more on anything else
Replies: >>221959
What would be an "immersive" way for a game to express that someone got banned, in a medieval setting?

For example the could be lore about a god that's banishes demon spawns who escaped from the underworld, and when someone gets banned there's a message "[god] sent [player] to oblivion".
Replies: >>221873
what is this garbage public fucking witch trials loser the term banhammer even comes from that me thinks
Replies: >>221875 >>221877
charge by mashing is retarded, what are you doing?
Replies: >>221959
What the fuck are you on about? The only thing I got was "public witch trials" and that gives too much attention to bots/cheaters.
Replies: >>221876
its public FUCKING witch trials idiot you know nothing burn
Replies: >>221877
draw a space marine
[Hide] (286.7KB, 749x449) Reverse
[Hide] (13.4KB, 420x583) Reverse
[Hide] (23.9KB, 420x583) Reverse
[Hide] (19.9KB, 420x583) Reverse
Working on some art for the yellow painting stuff and more variations of the convo NPC system.
The next levels in my game will be more of the Nortubel themed before going to a world ripping off a certain comic series.
Replies: >>221916
>niggers and dykes and blue hair
What the fuck are you doing anon?
I would strongly recommend not using this for serious image manipulation.  Even GIMP is better.  Paint.net is literally just MS Paint with a better UI, so things like text overlays and brushes will always be wonky at best.

>Currently I have 12 attacks spread over 3 buttons.
That's honestly probably way too many for a 2D action game where you'll be fighting hordes of enemies.  Even the complicated 2D games I've played only have about four or five attacks at most, not counting sub-weapons.

The issue with having more distinct attacks is that the more a player is afforded choice, the more he will opt to avoid the challenge entirely.  Worse yet, a dominant strategy may emerge unless you anticipate all the possible combinations a player can take and design levels around them.  A style meter and scoring system can help to force the player to use different attacks, but that may not be the best solution.

If you simply want the player to revel in gore then this is less of an issue but it is worth considering.

As >>221793 points out, Smash Bros is both easy to pick up and (relatively) deep because you can combine your movement and attacks in various ways.  There is technically only 1 attack button, but there can be huge differences between hitting with particular hitboxes or active frames, and depending on your trajectory the same air attack can cover a lot of different angles.

Having higher-risk attacks that do more is a good step, but I can't decipher your attack diagram for the life of me.  What does the quarter circle under "normal" mean, and what does the >> under that mean?
Replies: >>221959
>What the fuck are you doing anon?
I second this motion
Replies: >>222116
[Hide] (395.8KB, 2937x2178) Reverse
I'm making a fun game.
It wasn't .net 
I don't think it's that complex. Having dominant strategies isn't a big concern of mine, I just want to give the player a few movement choices with each having a slow but strong option and safe but weak alternative. Then there's movement attacks which are mostly to move around the map and to get away if you're stuck in a corner, and the extremely situational charged attacks. 
The hordes of enemies aren't going to be the main threat, or even a threat, and are just there to let the player try things out and to be a minor annoyance every once in a while. Most of the difficulty is going to come from the gimmicky bosses.
Replies: >>221963
[Hide] (23.9KB, 253x238) Reverse
>It wasn't .net
[Hide] (132.3KB, 735x524) Reverse
[Hide] (400.6KB, 1525x856) Reverse
[Hide] (4.7KB, 1096x211) Reverse
[Hide] (237.8KB, 518x435) Reverse
Worked on this model this week, its based on the Crawler from Breath of Fire, its around 800 triangles, most of the triangles are in the legs, and uses a 256x256 diffuse map for texture, the second image is how I imagine it will look in the game, I'm using FF4 remake and FF3 remake as reference to art style. 
I'm not satisfied with the final look and I have no idea how to improve it, from above it looks mediocre, maybe I should have used a photo of a real pillbug as a base reference than just the game sprite.
[Hide] (2.2MB, 1920x1080) Reverse
[Hide] (45.6KB, 850x453) Reverse
I'm trying to make that lava shader from >>221695, but it's pretty bad right now (this is the place when the instructor goes figure it out on your own, haha). The lava is supposed to be an emission shader, but if I set the strength low, it doesn't seem emission at all, and if I increase it until it looks like it is emitting color, I end up with uniform yellow. Second pic is what I have for the lava part. (The other parts are also not perfect, but they're not that bad).
Also, I had to switch to cycles because in eevee, every time I change anything in the nodes, it goes into "compiling shaders" for minutes, using a single CPU core, while cycles can render an acceptable image in seconds. Is it mesa's glsl compiler that is ubershit, or is it blender that generates uberbloat shaders?
[Hide] (96.2KB, 512x464) Reverse
[Hide] (277.4KB, 573x510) Reverse
>but if I set the strength low, it doesn't seem emission at all
I believe that is the normal behavior of an emission shader, you need to increase the strength of the emission shader to get that glowing look.
Replies: >>222082
>increase the strength of the emission shader to get that glowing look
But if I do that, my emitted color becomes uniform light yellow, no matter what color I set. At strength 15 it's already almost uniform yellow, and to get that real effect I need a strength of 100 or more.
Replies: >>222083
[Hide] (77.8KB, 1479x571) Reverse
>it goes into "compiling shaders" for minutes,
Try changing the 3D value in the noise texture to 2D.

See if this can help, use the BW noise as an input for both color and strength.
Replies: >>222090
The color ramp is wrong.
Move the two right sliders more to the left.
Replies: >>222090
It's obviously a mom-swapping game.
Replies: >>222086 >>222116
its birthing persons bigot
[Hide] (715KB, 903x574) Reverse
But then I get an even more yellow lava. If instead I move everything to the right (a lot), I get a weird pink color.
>Try changing the 3D value in the noise texture to 2D.
But then I either get constant values along the Z axis (if I leave the default object coordinate input) or discontinuations in the noise around the UV seams.
>use the BW noise as an input for both color and strength
Yeah, that seems to help
Viewport screenshot, because I'm on my crappy notebook now and it's already melting down from trying to raytrace this shit
Replies: >>222097
This is better. 
Replies: >>222121
I made the characters the way I wanted to make.
Even if the art is shit, there's technically an intention of mimicking anime-ish aesthetic or trying to add cute/waifu appeal in them.
(And I got a variety of characters to add in the game).
If things were different, they could've looked ugly in a "bodies and spaces" way.
I also want to mention that the girls still love a guy among them, so it could be a love triangle.
I still expect some PC freaks to get mad at something at some point.
They're older sisters to the 2 boys in the right.
It's more of a 2D companion game where you play as older siblings and must reach the exit with both characters.
There's also instances of playable/main characters ending up as NPC's in some levels.
Replies: >>222121
[Hide] (2.2MB, 1920x1080) Reverse
[Hide] (1.2MB, 1920x1080) Reverse
>mimicking anime-ish aesthetic or trying to add cute/waifu appeal in them
Not to criticize you, but right now it looks more calarts meet mspaint, than anime-ish. You should take lessons from the bunny-guy here.
Also those freckles(?), you should change their position and/or color, because right now it looks more like she's having stubble on her face...
>PC freaks
Why do you even care about them? No matter what you do, they won't be satisfied.

Hmm, I added that blackbody node, maybe it's better. And pic2 is what I'm trying to recreate, but I guess I'll leave it like this for the time being, I have like 20 hours of videos remaining to watch anyway.
Fun fact, on my crappy laptop with optimus, nvidia drivers only needs 5-10 seconds to compile the shaders instead of over a minute with amd/mesa. But on the other hand, they make blender crawl down to something like 2FPS, making it also pretty much unusable.
Replies: >>222132 >>222149
CalArts as in the Disney influence that John K pointed out or the Steven Universe/modern American cartoon one?
I also tend to trace over posed 3D models just because.
All in all, I want to get as much content as possible and then redo some art since I even collected a bunch of art tutorial images and have them tagged in my Hydrus collection.
The freckles do look off but I might some day fix them. I assume they look like a beard because of the lower positions, no?
>Why do you even care about them? No matter what you do, they won't be satisfied.
That I know exactly and is why I even brought it up in the first place.
Replies: >>222155 >>222459
>not satisfied and no idea how to improve it
What do you not like about it?  I think it looks fine enough for a single model, but the legs aren't that distinct so they might blend into the background.
>shader compile time
Its weird, when I had 8 gb ram, it took from 15 minutes to forever. But on 16 gb its below minute.

Play around with noise values. Dont use 0-1 color ramps, use something like 0-1-0.9, so top layer is not perfectly smooth. I would use math node to add voronoy and noise textures, and use "minimum" setting, so transition is smoother. 
Increase contrast and saturation of texture. Maybe decrease brightness.
Replies: >>222155
If you leave it at a size like that, you should increase the size of the important features, even if it's unrealistic. E.g. I'd definitely make its eyes bigger. And more contrast.
Don't know about the final shading/shadows/etc, but if you don't have them you should probably add some kind of outline, because right now it pretty much blends into the background.
>Steven Universe/modern American cartoon one
Probably that.
>when I had 8 gb ram, it took from 15 minutes to forever. But on 16 gb its below minute.
No, it's the driver. With nvidia's binary blob, it's 5-10 sec, with mesa (either on intel or amd) it's slow as shit. But I think it uses llvm, which is not exactly known for its speed.
Replies: >>222160
Okay, I've found a workaround, at least for amd cards. Run blender with MESA_LOADER_DRIVER_OVERRIDE=zink env var (you'll need a mesa build with zink support), because apparently the vulkan driver uses ACO to compile the shaders, not the retardedly slow LLVM. Why do you need to convert opengl calls to vulkan and why can't the opengl driver use the same glsl compiler as vulkan if they're in the same fucking package I don't know, but the open source graphics stack was always a mess.
Replies: >>222161
I spoke too soon, with zink the file selector in blender doesn't work. And there might be other bugs, sigh.
[Hide] (150.8KB, 492x943) Reverse
Spent the last two days texture painting this shit. I guess part of my soul died.
But my 15 years old cheap chink drawing tablet is definitely dying. if this goes on, I'll have to get a new one. Does getting one with a display worth the extra money? From what I've found on the internet, if I want something that works on linux, I should get a wacom, but they're fucken expensive. (There's also that digimend project which supports other tablets, but based on what I read, and also their github issues, it's more likely to be broken than not. There's also that opentabletdriver crap but it's a fucking garbage written in .net that also needs gtk3, so I'd avoid that abomination. Same goes for proprietary garbage supplied by huion and maybe others.)
Anon that looks amazing compared to my hilariously bad texture work.
Replies: >>222438
[Hide] (252.2KB, 820x674) Reverse
I am experimenting with using the cycles render after watching some tutorials however I can't get some of the textures to display properly, or at all.
Replies: >>222441
But I'm still just following a stupid tutorial, When I'm on my own, I end up with things like >>221713
Also I think something's off with the blend file, no material has a texture. Or did they just break backward compatibility again?
Also, holy shit, she has boobs?
Replies: >>222440
>no material has a texture.
Well fuck my life then. I have no idea how that happened since it shows all the textures and materials I set in blender.
[Hide] (35.1KB, 479x483) Reverse
Hey that looks breddy gud
Replies: >>222442
Except I can't give her a fucking face.
Make a horror game and she can be one of the spooky no face enemies that just glide towards you making weird sounds
Replies: >>222444
[Hide] (238.2KB, 789x296) Reverse
She'll be spookier once I figure out why textures aren't working right.
[Hide] (411.7KB, 1922x1064) Reverse
I think these days you need to use shader nodes and give it an image texture for base colors.
Replies: >>222452
[Hide] (197KB, 700x631) Reverse
Haha I figured it out. I was just missing a texture node, but the face and neck don't match despite being the same color.
Replies: >>222454
[Hide] (156.8KB, 643x601) Reverse
Hell yeah I figured it out. Let me know if there's any missing textures this time.
[Hide] (879.8KB, 624x1057) Reverse
>I also tend to trace over posed 3D
Why not trace over 2D since that's what you're going for? 
When you're horny enough even mu-onnas have fucking faces.
Replies: >>222556
I guess I'm not horny enough.

Also bumping my original question about tablets >>222385
Replies: >>222562
I've heard that tablets with display all have a problem with the glass making it so the point you press is slightly off from where it registers and it takes some getting used to, they're also more expensive as you've said. 
If you think your current tablet makes it harder for you to click where you want to it might be worth it to try a display tablet at a store if there's any around. Just walk in, say you're interested in display tablets but would like to try them out first, and don't feel pressured to buy it straight away.
Replies: >>222563
The biggest problem with my current one (other than it disconnects every 5-30 minutes when I have to replug it to start working again, and even when working it randomly not registers movement) is that there's a huge lag between I do something on the tablet and I see my mouse cursor updated. If you're drawing on a screen, there's obviously no delay between you move your hand and the time you see the pen move... but this is also the cheapest tablet I could buy 15 years ago, so I probably shouldn't base my expectations on that.
Also the pressure sensitivity is shit, I have to push it into the tablet like I want to tear through it to get >50% pressure readings...

>parallax effect from glass
Yeah, there's that, but it can be calibrated somewhat from what I read. Bigger problem is that some people say that with a display your hand can be in the way and and block out parts of the display, while without display you don't look at the tablet, so there's no such problem there. But I guess it also depends on whether you're sculpting or actually painting.
>at a store if there's any around. Just walk in, say you're interested in display tablets
Um, I don't think I've done anything like that in the last 10 years. 3DPD can be scary.
Replies: >>222566
>huge lag between I do something on the tablet and I see my mouse cursor updated
I have a huion giano and there's no lag at all unless I open the program before starting the driver, then it all goes to shit until I restart it. 
A display tablet is still a tablet, all it does is replicate the screen. If you have lag that's either on your computer or the input and an extra screen won't change anything. It won't fix the lag, it will just show the lag twice. 
>I have to push it into the tablet like I want to tear through it to get >50% pressure readings
Isn't there a way to change the pressure curves? I'm pretty sure mine does. 
Regardless, all problems you've said aren't inherent to drawing tablets and can also affect display tablets. You shouldn't buy a tablet with a screen unless you're buying it for the screen. 
>3DPD can be scary
Walk in and say "hey, I'm interested in these display tablets but I've never used one, can I try one out?". If they refuse just walk out, otherwise fuck around with it and don't draw huge tits or lolis in public. If you want suggestions say what you like and what you don't like, maybe you have a problem that can be fixed with software or another model (such as your pressure problem) and the clerk will inform you. Afterwards just thank the guy and say you want to think on it; if you're scared of getting pushed into buying it on your way out just do something that would insinuate you're going to return, like asking for a business card with the store's number or asking when they're open. That's it.
Replies: >>222572
[Hide] (357.8KB, 1920x1080) Reverse
[Hide] (219.2KB, 1920x1080) Reverse
[Hide] (177.6KB, 1920x1080) Reverse
Made some small progress on my PS4 emulator this week:
>Implemented instanced draws
It was fairly quick to add these as they only depended on a new packet type and on loading the instance index to a shader register. The programs themselves to the rest of the work.

Some other work went to the graphics library used by the programs:
>Wrote a very basic GCN assembler
It has only some microcodes available but it was easy to build as it reuses a lot of definitions found in the library's disassembler. It was a requirement for the next point.
>Improved fetch shader builder
The original looked very messy in some sections as it was reverse engineered.
Some problems when using instanced vertex inputs were spotted too.
I'm glad that I wrote tests for it before, otherwise I could have had a bad time debugging this on the console.

I also got sidetracked and wrote a plugin for the PS4 to change a game's language independently of the system's.
Replies: >>222576 >>222577
>either on your computer or the input
I'd guess it's the tablet, as the mouse or keyboard doesn't have that lag. Fuck, even my bluetooth gamepad has less lag than this.
>Isn't there a way to change the pressure curves?
There should be, I'll have to look into this evdev/xinput mess. At worst, I can patch my kernel driver, khm.
>Walk in
Yeah, second problem I also need to find a shop that sells them. Being in a small yurop country, they sell absolutely nothing in the shops other than what peasants under IQ 50 need, everything else is only available in webshops. I only found a single shop that actually has wacom tablets in stock which is not a webshop, on the other side of the city.
Replies: >>222573
[Hide] (6.6KB, 552x90) Reverse
[Hide] (31.2KB, 806x516) Reverse
I would not recommend a screen for few reasons
>for some odd reason, majority of screen tablets aren't even close to color accurate
>wrecks your posture, namely your neck
compatibility is fine but you might get tripped up with trying to get the tablet working properly with multiple monitors, look into digimend for non-wacom tablets

some art programs allow you to change the pressure curve, see pic-related for blender & krita respectively
Replies: >>222615
[Hide] (146.1KB, 320x228) Reverse
WOHOHOAH! Nice anal beads bro!
Replies: >>223426
Does this mean I can play gravity rush 2 on your emulator soon?
Replies: >>223426
>color accurate
<implying my current monitors are anything but accurate
>wrecks your posture
I think I'm already past that point. But some tablets can be used upright, are they any better?
>look into digimend
I did that, but what I read about it was pretty bad, and I'd rather not buy an expensive paper weight.
Replies: >>222619
Do you really need it? Mouse and keyboard are really good to begin with, its hard to upgrade from them. From what I seen, most of tablet users use it for extremely small details, such as wrinkles, and it seems they spend weeks working on a single model, not something I would bother doing. 
And you can probably bind "pressure" to mouse wheel or something similar anyway. As the other posters said, just try it out somewhere.
[Hide] (4.7KB, 1096x211) Reverse
[Hide] (88.4KB, 491x366) Reverse
[Hide] (119.5KB, 526x392) Reverse
[Hide] (127KB, 578x433) Reverse
[Hide] (32.2KB, 247x272) Reverse
My second attempt in making the crawler from BoF1, this time I started with a black white texture and applied the colors using a rgb map and color ramp, even tried to use a brush to paint the black and white texture, and I tried to follow the advice anons gave me, I think this design looks better than the previous one but  I'm not sure if I managed to accomplish what I'm wanted, now I need to work on the second form.
What do you guys of this new model ?
Looks very good anon. What are you making it for again?
Replies: >>224081
colors look good as is, but not as accurate to the pixel art imo
the bug looks way more metallic on the sprite
Replies: >>222888 >>224081
Nice for organic, but as >>222885
said, colors values need to be more harshly contrasting to capture the metallic look of the original.
Replies: >>224081
[Hide] (617.2KB, 942x451) Reverse
Looks nice, but consider making legs bigger, with just black silhouette/outline. Or just make them bigger. 
Also their number. Consider it like fingers on simpsons characters, make as many as long as they look good. It is a big problem when you want details, but they simply dont look good on such small scale. So just didnt try to "design" a creature. For example, you dont need to use the same model, and same number of segments when it wraps into a ball, just use same segment any number of times, as long as it looks like what you need. 
I know you go for stylistic look, but this is an example of bug with better proportions. 
I didnt bother to properly unwrap, and lights are for texture baking, etc etc, this is the super fast and easy setup.
Replies: >>224081
Can I vent about blender again?
Trying to figure out why the transform values don't make sense with a parented object. First I tried to reparent some objects, but it produced weird results. Upon closer inspection I also noticed two objects parented to the same object, both are around the same position, one has z=0.7, the other has z=3.1. What the fuck? After some digging out, blender has this fantastic "feature" called parent inverse matrix, where it puts a matrix between the parent object's transform and the object transform, which is not exposed anywhere in the fucking GUI. All you can do is clearing them (setting the matrix back to identity), but in that case you'll have to manually fix the fucked up transform afterward. There's an apply parent inverse option, which according to the documentation:
>Applies the object’s Parent Inverse transform to the object data.
But that's not what it does. If you look at blender's source code, it says
>Use parent's world transform as the child's origin.
So it won't apply the inverse matrix and set it to identity, instead it'll fuck up even more. Just because some retard at blender can't think in local coordinates, so they come up with this abomination, where you end up with a frankenstein system that's neither global nor local coordinates, where the numbers displayed on the GUI doesn't fucking mean anything.

But anyway, now that you know it, if you're bitten by this, get this addon:
this will allow you to get rid of these retarded matrices without having to manually recreate whatever transformations you had (it's bound to Shift-Alt-P by default).
If you want to reparent something to something else, you'll need Ctrl-P, select the simple parent option, then select the objects you parented again and Alt-P, clear parent inverse. There's an Object (Without Inverse) in the parent menu, but that doesn't fucking work, what did you even expect?
I really hope whoever came up with this nonsense will be randomly shot by some allakhu akhbar.
[Hide] (462.8KB, 1600x900) Reverse
I'm working on Teagan as we speak, using all the techniques from my previous model as well as using the tutorials provided. It's going fast and I hope to be done somewhere around the fifteenth of this month.
Replies: >>223418
Tits are too big, Teagan may have respectable chesticles but she's too fit to have the floppy khazar milkers of Alto.
Replies: >>223429
[Hide] (12.7KB, 1920x1080) Reverse
[Hide] (32.2KB, 1920x1080) Reverse
Did some very small changes to my PS4 emulator this week:
>Initial work on custom primitive types (rectlist, quads, ...) conversion
Currently it only converting rectangle lists, and does it incorrectly. It's all setup on CPU, but might go with geometry shaders like GRVK (Mantle to Vulkan implementation) does since they look like a much better solution.
Separately from the emulator, I submitted a patch to MiraHEN to fix logging after suspending and resuming a PS4.
Slow week overall, hopefully the next will be more productive. Also, I missed that the emulator turned one year old last October. Here's to another year, this time with actual games to play.
No, and won't be for a while I reckon, if at all.
I'll probably target GR1's remaster first since it's more simple, and even before that some advanced 2D game like Dragon's Crown Pro.
There's too a bunch of stuff that needs to be done, like getting proper GPU memory emulation. Only AMD cards can currently run the emulator because of that.
They only look that big because of the angle. I'll probably reduce their size later.
Replies: >>223442
Good point. Post a 3/4ths view so we can decide for ourselves.
Also separate breasts may not be the best choice if you're not intending to reuse these models for lewd purposes, Teagan is generally shown wearing a turtleneck sweater, not boob socks. Since you said you're planning on using these models in JKA I don't know how well the Quake 3 engine will handle stacking another mesh on top of the boobs. Ignore me if you're planning a bikini skin or something.
Replies: >>223448
[Hide] (131.1KB, 640x726) Reverse
Well okay if that's what you want.
Replies: >>223508 >>223526
[Hide] (270.6KB, 865x875) Reverse
I think he meant something like this. Also put more geometry into shoulders and try to avoid using non-quads.
Replies: >>223517 >>223526
[Hide] (95.6KB, 640x638) Reverse
looks like my ass
Replies: >>223518 >>223520
is your ass ok you should go to a doctor
[Hide] (713.7KB, 1123x861) Reverse
I actually made an ass, but never finished the rest.
That's not as big as I thought. I'd reduce them by at most 10% and push them together slightly for proper sweater puppies.
This is a little too small.
[Hide] (204.1KB, 1920x1080) Reverse
[Hide] (424.3KB, 1918x1060) Reverse
>3D Software Developer and Researcher, Technical Artist
I should have gone first with this guy instead of that other who told plain wrong info about sRGB and blend modes and maybe others. Learned more things about bricks than I ever wanted.
... On the other hand, I like how in the intro he said try to keep the detail value of noise textures as low as possible because they're slow, then later goes on using insanely high values where you only see any difference if you look at the wall from 5 mm. I got to the point of having a thumbnail sized viewport, so eevee can muster about 0.5FPS and it only freezes my whole desktop for multiple seconds every 10 clicks, not every second.
Replies: >>223606 >>223610
That shader graph looks like a starship.
[Hide] (1.1MB, 2048x2048) Reverse
[Hide] (3MB, 2048x2048) Reverse
[Hide] (1.2MB, 2048x2048) Reverse
[Hide] (301.3KB, 2048x2048) Reverse
>try to keep the detail value of noise textures as low as possible because they're slow
90% sure its misconception. When you use any procedural node, you basically use a formula once per pixel/bounce, resolution is just a variable. You dont actually generate a giant 32*32*32 images based on it. Having low noise resolution is as good for performance as using xyz coordinates close to 0, its the same amount of calculations. Even if they were different it would be difference in just a few cpu cycles at worst, you still do the same calculations.
>If the hardware is (or is like) x86 with legacy x87 math, float and double are both extended (for free) to an internal 80-bit format, so both have the same performance (except for cache footprint / memory bandwidth)
While gpu instructions are different, I doubt gpu is any different. 

>insanely high
You use as high of a value as you need for the size and roughness of noise. 
>only visible from extremely close
It becomes a problem when you have to render in higher resolution, to keep the details without turning them into a mush, due to denoiser/noise. You can have the most clean and perfect pattern, but it would look like smeared shit, because resolution is too low (or object is too small). Its a big problem, since you want to have these details, but at the same time, they are not only invisible, they decrease overall quality. This is why pixel art and insane texture resolutions are popular, fitting a lot of details into medium resolution is harder than fitting a lot of details into high resolution or simply using less details. 
Its not a realtime rendering engine, it uses minutes per frame, not frames per second. 
>wrong about things
Yeah, even people who made these parts are wrong about them, its a mess. Often parts were broken, so people made addons to fix them, but later they were fixed, but memory "this is broken" remained. You dont have to worry about most of it.
>that node setup
Unless you dong something extremely realistic, or have extremely good PC, dont bother. In many cases, extremely elaborate designs look worse and dont actually improve anything. 
But consider this an example of how to do individual things. Blender materials are designed to be exhaustive, you make a single material and you dont need to ever make it again, since it would work in any situation. In the end you have a single node with a bunch of sliders which do everything you need, so you save time by making it, But if you only needed it once, you wasted time on making it (but it is fun). 
Also also, these days you might want to use geometry nodes setup for such things. While material setup and geometry nodes setup might be identical in many ways, using materials to simulate geometry sometimes goes too far. Often its easier to make a single brick, make a material with damage on it, and use a geometry node setup to place them around. It is a balancing between gpu memory, ram, cpu power, etc. If you use displacement for bricks, they would need more vertices, and if you have a million bricks it would be impossible to render (even if they are instanced, assuming its possible with material deformations). But if you use geometry nodes and deform bricks in it, it would be easier to render. In fact you can use a setup to bake textures, and whole realistic rendering would take a few minutes instead of days, despite having a (literal) million bricks.

Oh, and btw, I said it before, but a lot of these guides are just commercials. You can do what they did, but they already did it, and it only costs 15$ to get this and a few dozen more materials, without wasting any of your time on making it yourself. So they use most elaborate and "expensive" designs ever, for showcase. If you are making some effects for the movie you might use such setup, but in reality, a lot of people who actually make effects for movie use "just use a photo, lol" approach. For example making grass. You can make elaborate generator, make a perfectly realistic and unique grass, with virtually infinite resolution. Or you can use picrelated, same as everyone else (cc0). Making your own grass generator might be fun, but it will be a waste of time, since picrelated exists.
Replies: >>223635
[Hide] (1022.4KB, 1150x494) Reverse
>resolution is just a variable
I was talking about the detail parameter, not the scale. (If I can believe this guy), this noise texture works by generating multiple noise patterns with different scale and combining them and detail defines how many layers you want.
>float and double are both extended (for free) to an internal 80-bit format
That's no longer true since you have SSE, i.e something like early 00s; and on GPUs doubles are ridiculously slow, especially on consumer novidia GPUs (I head they slow them down on purpose, so people who want to do GPGPU shit have to buy the overpriced datacenter GPUs). Nobody does double precision calculations in computer graphics, heck, some CPUs/GPUs even natively support half floats (16-bit) for cases where space saving is more important than the extra precision.
>Its not a realtime rendering engine
Quoting straight from blender's manual:
<Eevee is Blender’s realtime render engine built using OpenGL focused on speed and interactivity while achieving the goal of rendering PBR materials.
>geometry shaders
Yeah, I'd probably go with that if I were on my own. Especially if you look at the bricks from the side it's painfully visible that it's just a height map with a 2D texture projected over it.
This is actually a course that would have costed my $45 if I wouldn't have bought it from torrent shop, so a without any correction for deptht least it's not a commercial, and you only get a single shader out of it. But still he seems to overdo it.
>just a photo lol
Photos can give you colors, but to make them beliveable you still need some roughness/normal/metallic/whatever map if they're not constant along the surface. I've seen people converting the image to grayscale and adding some ramps to come up with these textures, they work sometimes, but I dunno about more complex cases. I heard that there's AI for this too, but never looked into that.
Replies: >>223638
[Hide] (172KB, 1220x860) Reverse
>I was talking about the detail parameter, not the scale.
It too is just a variable. 
><Eevee is Blender’s realtime render engine built using OpenGL focused on speed and interactivity while achieving the goal of rendering PBR materials.
Well it does render in real time, just slowly. 
>he overdoes it
But, as I said, it is "complete" method. Do it once, and use it forever. And it is universal skill. 
No, people literally take a photo of a building/object, and import is as a plane, tweak roughness a bit and use it in movie. As you said, no one is going to notice a difference, and it looks realistic because it is a real photo. I mean, blender has IOR slider on pbr shader. Every matherial has it, and to be 100% realistic you need to adjust it, but no one does it, because its impossible to notice and "just use random/default value" works every time. 
>just add some ramps
Yeah, it is how most of seamless textures are made. You can be a bit fancier than that, but it is a basic principle. Hires photo, into seamless box, into a bunch of maps.
Replies: >>223641
>just a variable
Yes, a variable that defines how many times you need to execute a loop.
>it does render in real time, just slowly
The problem with that is it runs in the same opengl context as blender, so if eevee slows down, your whole blender starts lagging, while with cycles only the viewport is affected (and even that's only partly, outlines and gizmos are still updated realtime).
Replies: >>223643
[Hide] (1.5MB, 1920x1081) Reverse
>14.33 seconds for 16x detail
>14.41 seconds for 16x detail
>13.82 seconds for 8x detail
>13.26 seconds for 0x detail
For this 8x noise into noise setup. 1080p 32 samples, nothing to bounce off. So, while it apparently has an effect, its very marginal, I doubt it works the way its explained, on actual hardware.
Replies: >>223700
[Hide] (41.7KB, 600x315) Reverse
Does anyone here have any idea how to create different LOD for models?
I've got a fairly high resolution model I want to shrink down the vertex count from.
Does creating LODs use nodes? I hate nodes
Hmm, I expected bigger difference. But anyway, I looked it up.
Noise texture with the detail parameter here, calling fractal_noise:
Here detail is called octaves, but here is the for loop:
Which is like the meat of the shader (it won't go into that big if when you have an integer detail). That texture stuff will call this again twice to get green and blue channels of the color output, so it doesn't really do anything besides that loop.
How long it takes to render the same thing without all these noise nodes, just plugging a constant color into the shader?
Replies: >>223707
>try x
Not exactly the same, but jut a box with default shader, stretched to cover whole screen takes 9.26 seconds to render. 
Either way, you should not optimize it, because optimizing it wastes more time than saves. And doing the exactly same thing twice in a row takes less than double time, since all the data is already fetched, its already in registers, and everything is "split" between cores.
Demo day this Saturday?
[Hide] (1.6MB, 1024x600, 00:23)
[Hide] (816.9KB, 00:54)
[Hide] (75.6KB, 272x182, 00:03)
Got off my ass and finally did some sprites I needed to test stuff out. I think the combat might be fun but I'm really not feeling the hopping. 
I also made a couple new songs, including this one which I might change. I like it and it sounds good when playing it unplugged but with distortion on the middle part ends up sounding too much like E1M1. Also, ignore the fuck up, I accidentally the drums and only noticed after exporting it and nuking the raw guitar. 
It's surprisingly easy to stumble upon popular riffs when trying to find something that sounds good. I even found this weird magical tune. When I played it space and time distorted and it spawned a cease and desist from nintendo.
As for the graphics I might drop Grafx2. It's good and all but the technical side is a mess. It's extremely slow during certain operations, freezes occasionally, and I've even had it crash it multiple times, most of which came from trying to open files it had just exported. Trying to do serious work while there's a chance the program might just eat shit isn't fun, trust me, I'm familiar with the concept. 
My current sprite setup is clunky, but it could work well with ms paint. Just draw bases on CSP, export png, use gimp to stack them all together and fix the sizes, export as gif with replacing layers, use ffmpeg to turn the gif into multiple pngs, then use GlueIT to turn the pngs into a sprite. After that I could reduce the opacity or something and use paint to draw the final sprite over it. 
Jank as fuck, but not even Godot is free from the jank because this god forsaken program has functions that work differently in the program and in release builds. 
Anyway, builds because next update might take a while.
Windows files.catbox.moe/b8ddbe.zip  Linux https://files.catbox.moe/94zmav.zip
Now I need to force myself to finish the remaining test sprites to then test out the combat and movement systems. Should be quick once I get into it but might take a while to get into it.
Replies: >>223763 >>223778
[Hide] (98.2KB, 376x376) Reverse
What software you're using in your magic_song.mp4 sample?
Replies: >>223771
Tuxguitar. It's a FLOSS version of Guitar Pro.
[Hide] (148.6KB, 1920x1080, 00:05)
something is fucked up with physics (did you do everything in process without using delta time? - which you should be doing that shit in physics instead and arguably all of it in physics ) and some of the state stuff like jump > land > hi jump > fall attack is different than just doing a fall attack
i also think it's weird how the motion inputs for attacking has to always hold forward to have the move come out and on top of that I recommend flushing out SOCD inputs, holding left and right always makes you move right

excuse the shit video but demonstrates my point
Replies: >>223781 >>223783
[Hide] (342.8KB, 1024x600, 00:04)
[Hide] (4.8KB, 214x130) Reverse
That's weird.
>did you do everything in process without using delta time? 
I think I found the error. The fall calculation is based on a curve, the end result does get multiplied by delta but the value you get from the curve is still dependent on when the function is called and can cause some oddity if the frames aren't consistent. physics_process should fix that. 
I don't know why the wiki barely mentions the function exists and even introduces you to physics processing with the _process function instead. 
https://files.catbox.moe/cre5cl.zip https://files.catbox.moe/n9twra.zip
>i also think it's weird how the motion inputs for attacking has to always hold forward to have the move come out 
You mean with the quarter circle thing? It just checks the order you pressed the keys and if there wasn't more than half a second between presses or other keys in between. You don't actually have to hold forward and can even wait a while before attacking. 
>recommend flushing out SOCD inputs, holding left and right always makes you move right
I used an if and elif because I didn't consider anyone would hold two keys but  that makes sense. 
>excuse the shit video
Protip, if you use a rectangle to outline the part of the video you want in gimp if gives you the numbers you need in almost the exact order ffmpeg's crop .
Replies: >>223792 >>223799
>some of the state stuff like jump > land > hi jump > fall attack is different than just doing a fall attack
Also, what do you mean by that? Does the animations/text fuck up or does the attack behave differently?
>I didn't consider anyone would hold two keys
But that's actually very common. When you suddenly have to switch from going left to right, you can easily end up pressing both keys for a short time, and if the game does something weird in that case, it's annoying.
Replies: >>223847
[Hide] (1.5MB, 1920x1080, 00:14)
[Hide] (1.8MB, 1920x1080, 00:15)
>You mean with the quarter circle thing? It just checks the order you pressed the keys and if there wasn't more than half a second between presses or other keys in between. You don't actually have to hold forward and can even wait a while before attacking. 
disregard that, i just realized i'm retarded and found out that doing quarter circles on keyboard is a bit more precise than i remembered

>Also, what do you mean by that? Does the animations/text fuck up or does the attack behave differently?
the attack behaves differently, your falling attack gets interrupted and the hitbox sticks out forever until you land.\
the first vid related shows it off, the last attempt shows off a hijump variation

also more notes with the fixed bulid
- being able to turn mid air, is kind of weird. along with able to turn on the spot while running
-double tapping forward/backwards quickly while in the air halts/locks all horizontal movement
- moving forward after light attack gives a very fast jab, same with the other attack (see 2nd vid related)
Replies: >>223847
[Hide] (879.3KB, 1327x2096) Reverse
[Hide] (6.1MB, 4000x2893) Reverse
Spoiler File
(994KB, 1040x1300) Reverse
Checked out this material maker. In the online material collection, they have procedural
>wall made of old cellphones
>shader that turns a sphere into a snowman or a teapot
>realistic bird poop
>screensaver used by old dvd players
>server rack with ethernet cables
>wall of ticket machines
>crawling scarab beetles

Also, tangentially related question, about art styles. There's realistic, when it tries to be as realistic as possible, there's that anime style with contours and flat shading, sometimes even going as far as only using flat tints, then there's this thing in picrel, somewhere between the two, it's clearly not trying to be realistic, but it has gradients on the shading (even if there are some hard edges), sometimes with practically no outline. Is there a name for this? Also, any example not involving people?
Replies: >>223860
Skill issue. 
It's neutral now.
>doing quarter circles on keyboard is a bit more precise than i remembered
Just hold down, press a direction, then release down. 
>your falling attack gets interrupted and the hitbox sticks out forever until you land
I have an animation node for attacks and one for movement for performance reasons. Turns out if you attack at just the right time it plays the stop animation before the attack happens, and I forgot to manually stop the attack afterwards so it just plays and stays active until you tell it to stop again. That might have fucked me later.
>being able to turn mid air, is kind of weird. along with able to turn on the spot while running
I am going to add some sort of inertia for running eventually, and maybe for jumping too. 
>double tapping forward/backwards quickly while in the air halts/locks all horizontal movement
That was intentional at some point, but I forgot to undo it.
>moving forward after light attack gives a very fast jab
That's retarded, I'm probably going to keep it for the jab. 
https://files.catbox.moe/zze32g.zip https://files.catbox.moe/72rw3r.zip
Replies: >>224086
Someone should correct me, I think that is normal shading in a stylized character,
Replies: >>223872
[Hide] (216.4KB, 747x975) Reverse
Are you sure? Especially the highlights from the second pic, I don't think you'll get anything like that on a normal body. But true, most of the shading tries to be realistic, or more like, as realistic as you can get with a 2D drawing software without too much hassle.
Replies: >>224086
[Hide] (383.7KB, 883x603) Reverse
[Hide] (402.7KB, 1082x606) Reverse
[Hide] (336.6KB, 829x502) Reverse
[Hide] (313.6KB, 830x496) Reverse
[Hide] (301.7KB, 855x492) Reverse
Ball form for the bug, I've increased the size of the legs for the normal form, I've added two outlines for the model, the silhouette has transparency value, I think the model look much better, I'm experimenting with the roughness, metallic and specular values in the BSDF node using the BW texture as an input.
I know texture are usually stored in sizes of power of 2, 1024x1024, 512x512, etc, is there problem if store the texture in a rectangle, 128x256, 512x128 ?

Is this better ?

you get a gold star

you get a grey cloud

1 gold star removed
Replies: >>224092
POT rectangles should be fine. To be honest, probably even NPOT, any GPU made in the last 10 years shouldn't have a big problem with them, but still, if you can make them POT, make them POT (mip-mapping is weird with NPOT textures).

Also, make sure to check the model in whatever engine you're going to use it, PBR is supposed to be a standard, but there can be differences. Also, as you can see from your last 3 pics, lighting influences it a lot, try to find something that will look good in the environments that will be used in the game. This could also mean putting the model away, and revisiting the question when you have an (almost) final level where you can test it.
>Is this better?
Somehow I find the green parts of the bug the best in the city image (just a tiny bit bright), while the rest is better with the studio lights. I guess you can go with high metallic and low roughness values, these little bugs can be shiny.
Be careful with specular values though: https://libreddit.kavin.rocks/r/blender/comments/hff5hw/should_i_change_the_principled_bsdf_specular/

you fuck off
Replies: >>224138
Looks fine to me. Good place to stop. But try changing texture interpolation to "closest" for pixelated view, it might look better. 
>non square textures
While they are perfectly fine, you should consider packing uv islands into a square anyway. 
And since you practicing, you might want to finish it. Game models dont have mirror modifier, and often they dont have mirrored textures, so consider applying mirror, and changing texture to include both sides. 
You would also need to animate it, so think about it. 
Single texture seem to be enough in this case.
[Hide] (2.8MB, 1600x962) Reverse
[Hide] (193.5KB, 1024x1024) Reverse
Does it look shit or does it look shit?
I've been looking at this for hours and it's still shit, the voronoi pattern is still too obvious and they're too uniformly sized, but if I increase randomness, they became blobs. (There's no background behind the paint yet).
I have no idea how do people come up with things like this. https://www.materialmaker.org/material?id=449
Do I have to sacrifice virgins and drink their blood or what?
Replies: >>224219 >>224257
Will there be a 11/11 demo day thread here or is that verboten?
Replies: >>224209 >>224210
Why would it be verboten?
there were demo threads made here before so yes as long as someone makes it
a good amount of guesswork and a borrowing other people's tips & tricks, on the same sort of line like ai prompts
The small cracks in the second pic look fine. It is just that the large cracks are too undefined and blurry. 
>There's no background behind the paint yet
I think that really is the major drawback right now. If you just added  that extra layer it would add the depth needed rather then the void the user is currently seeing.
Replies: >>224317 >>224327
[Hide] (4.2MB, 2048x2048) Reverse
[Hide] (2.2MB, 1614x824) Reverse
Yeah, I didn't want to create a really hard edge by doing an a>b kind of comparison, so instead I map a really small range to 0..1 clamped, but depending on the underlying noise texture at some places there's still too much transition going on.
Anyway, here's try #2, with the wood texture stolen from https://www.materialmaker.org/material?id=771 but tbh I'd need something a bit more worn down

One thing I should still figure out is that since this is a door, places where it is usually used (around the handle, and somewhere around the center where people usually push the door) should be more dirty/worn. In blender I'd probably go with texture painting a grayscale texture, and use that somewhere to control the cracks/etc, but since I made this in material maker, that won't work. One workaround is to simply export that texture and throw it into material maker, but in that case I lose the realtime preview ability. Alternatively I could convert this shit back to blender, but those buffer nodes are fucking useful, I can put more than 5 nodes into the graph without slowing my computer down to crawling speeds. Custom GLSL nodes are also awesome, even though I only made a trivial modification to one built-in node so far
Replies: >>224327
[Hide] (1.1MB, 1920x1081) Reverse
Looks fine. But try more musgrave texture, it looks the most organic, but sometimes its too fancy. Setup in picrelated, move sliders around, offset in musgrave is the most impactful. 
>material maker
Use blender.
>Yeah, I didn't want to create a really hard edge 
Yeah, I wanted to give this advice, I even started rendering an example, but blender crashed. 
>places where it is usually used (around the handle, and somewhere around the center where people usually push the door) should be more dirty/worn
Use ambient occlusion and geometry-pointiness for this. Assuming you dont have and older radeon videocard, just plug them in, change to rendered view, and adjust them as you need. Personally I just bake them, because I have an older radeon gpu. But sometimes its easier to just pain the dirt map.
>steal tricks from others
Basically this.
But you also need to consider everything as information, which you can use for any purpose. Any texture can be used as a vector, most of vectors can be split into individual data, etc.
[Hide] (118.4KB, 1920x1080) Reverse
[Hide] (22KB, 1920x1080) Reverse
Got some work done on stuff for my PS4 emulator:
>Implemented indirect drawing on graphics library
This wasn't too complicated: some new packets for drawing and setting resources were added, along with a structure for the indirect data, which was taken directly from Vulkan.
It's interesting how vendors agreed on the indirect data structures' layouts to be consumed by GPUs.
I'll see if I can get a more interesting indirect sample working later.
>Updated PS4 frame dumper to latest changes in the graphics library
It wasn't updated since July and some stuff is different now.
Old hacky way of dumping vertex buffers is gone and they're now treated as regular buffers now that the emulator supports them better.
Draws are no longer executed separately since that breaks some games.
The emulator isn't replaying new frames right yet since it's not replacing pointers in pointer/descriptor tables with new ones, which are a new feature.
Also the primitive changes from the other week were put on the side since that's more complicated than I thought.
[Hide] (1.1MB, 229x300) Reverse
sword game guy here, i made some improvements to the demo i posted a couple months ago
>sword now flies in the direction it's supposed to
>sword now gets stuck in the ground easier
>the birds move left and right + edge detection (still flightless)
still no win/loss condition and still very easy to get stuck/out of bounds, i hope to post more of these updates so you can motivate (bully) me into making it better.
Replies: >>224758
What in the fuck are the controls for this game.  I can sometimes get the sword to appear in the air and then I can aim and shoot, but half the time I'm just clicking random keys and hoping it'll go back to that state.  Perhaps an explanation?
Replies: >>224759 >>224764
Right I finally figured it out.  You have to right click on the exact part of the hilt to reset it, and you left click on the hilt to aim and shoot.  My first recommendation is to make the resetting easier - just a right click would probably be good.  I imagine once you have actual enemies that just having the controls be right click to reset and left click anywhere on the screen to aim would be both more intuitive and be necessary.
Replies: >>224764
that's my fault, when i'm testing the game I have hitboxes viewable so i can see where the hilt is to right click on it. i plan on making it so you can reset and launch the sword from anywhere instead of clicking on it since it gets very tedious very fast.
Raylib isn't engine dev, it's a framework.  I would know, since I use Raylib.  Autistic semantics aside, scope, I would assume.  A game engine would have a whole slew of features, including physics simulation, audio, 3d/2d support, AI, (optional) scripting/modding support, etc.
Replies: >>225950
[Hide] (159KB, 686x546) Reverse
[Hide] (148.4KB, 686x603) Reverse
[Hide] (155.5KB, 711x623) Reverse
Just so you know I'm not dead, here is the progress with Teags so far. The hair is taking quite a long time, but I'm making progress little by little. The hair is the most difficult part, once that's done I should be able to complete it pretty quickly.
why is she grey, is teagan an ayy lmao
Replies: >>225202
the symmetrical hair in the middle has to go
[Hide] (105.8KB, 1912x1035) Reverse
Here's what got in my PS4 emulator this week:
>Implemented indirect draws
Some stuff was rewritten to allow indirect and index buffers to be passed around better, but fortunately it wasn't too complex since you can pass indirect data from a game directly to Vulkan as it shares the same indirect data structures as GNM.
Image related has the indirect example running on PC. There would be some text at the top left if it wasn't broken due to swizzling issues.
>Shader decoder now parses MTBUF instructions
This microcode only has 5 instructions and is very similar MUBUF, except that it specifies the buffer's format explicitly. Kind of wish all of GCN did that since SPIRV wants to know most formats at compile time.
Only 3 microcodes remain to be decoded out of 12.
>Index auto draws now use vkCmdDraw
Despise its name index auto draws are essentially draw calls with a vertex count, so this should optimize the emulator a little.
Replies: >>225348
I managed to create a function that can print a string variable on the screen one character at a time! And that's not all: the funtion also trims the string and starts printing it on a location that's defined by the parameters. I'm really feeling myself.
are you the only developer of this?
Replies: >>226196
Spoiler File
(99.6KB, 635x1104) Reverse
I guess it's kind of a conflict between philosophical vs marketing definitions. It could defined as "Reduces all relevant asset pipe-lining to 2-3 steps. Handles all memory creation and destruction automatically."
Replies: >>226041 >>226071
Not sure if the last point is correct, since that sounds more like a language feature to me.  It's just that most engine languages are object oriented, and memory handling is a feature of object oriented languages (unlike C, which is imperative, and does not handle memory for you). Asset pipelining is a definite yes, for sure.  Most people also think you have to have an editor in order to be a fully fledged engine, though I wouldn't exactly call it necessary.  Just not nearly as accessible beyond your personal team, I would suppose.
Replies: >>226603
[Hide] (216.4KB, 1905x857) Reverse
[Hide] (4.3MB, 2048x2048) Reverse
[Hide] (593.1KB, 473x880) Reverse
I'm still alive, I think I made a ridiculously overengineered tree texture. Anyway, I'm not going to fuck around with it more, I had enough of it. Now I just have to figure out how to warp my 2d paint overlay onto 3d without any obvious seams where it shouldn't be.
Replies: >>226444
>that image
Just make real brownies; these take like dogshit.
Replies: >>226087
would calling them depression brownies make it more appealing to you?
[Hide] (1.5MB, 1024x600, 00:06)
Progress has been slow.  Just the thought of having to make sprites makes me not feel like working on the video.  There's now a tutorial area and that's about it. 
I recorded a couple more songs with my ghetto setup but hopefully I'll have an audio mixer within the week and that will allow me to hear the PC audio without my guitar having retarded delay. Once I do I'll re-record the current tracks because some are pretty fucky, including the new ones. 
And on the subject of songs, I wonder how common it is to make an asset, realize it's plagiarism, and then treat it as an easter egg to have an excuse to keep it. 
Here's the build, next update when I feel like it. 
Win https://files.catbox.moe/opqb8o.zip | Linux https://files.catbox.moe/dtaxcl.zip
Replies: >>226153 >>226165
Fuck, I wanted to add some more text like "the game has 18 levels and 6 cutscenes in total" but still.
Controls are not too bad aside from the kid getting stuck on ladders. Please let players fast forward cutscenes though, you shouldn't have to pick between slow story or no story.
Replies: >>226193
>Controls are not too bad aside from the kid getting stuck on ladders
I tried making the info more explicit on the yellow books from the tutorial levels.
>Please let players fast forward cutscenes though, you shouldn't have to pick between slow story or no story.
Part of why some stuff is slow is because someone else complained shit moved too fast.
I'd love to figure out a way to adjust speed, if it could somehow work in the structure I came up with and not break anything at all.
Because it's still my first game and me being an ideas guy that just happens to put some work.
Replies: >>226207
[Hide] (226.9KB, 1920x1088) Reverse
[Hide] (596.7KB, 1920x1080) Reverse
Worked on a few items in my PS4 emulator this week:
>Implemented image swizzling
Every shader image access gets handled by a switch that determines what value you get (either an image channel or a constant).
>Updated frame replayer to frame dumper's changes
Dumper now outputs some data differently so this was needed to make the replayer (mostly) work again.
>Frame replayer image converter now outputs PNGs instead of QOIs
This was done because QOI only supports RGB(A)8 formats.

On the side I prototyped a Reclass plugin for MiraHEN, which could help with the emulator.

>Part of why some stuff is slow is because someone else complained shit moved too fast.
I thought that might be the case. 
>I'd love to figure out a way to adjust speed
Are you using animations? If you are you can adjust the speed of the animation player from a script and make it go faster while the player is holding ctr or something.
[Hide] (5.1MB, 2048x2048) Reverse
[Hide] (530.8KB, 418x749) Reverse
Door failed successfully. But still looks like shit.
Only included that last point because in raylib you have to manually destroy assets before exiting the program. You have to make a pipeline that handles that if you use RL.
Replies: >>226894
Do anyone know some tool/anything that can be used to convert blend files to something that at least attempts to look like something that is compatible with version control systems and not just a big stinky pile of binary garbage?
I really like how materialmaker's files are just pretty printed json files that are more or less readable even if you don't know anything about its file format, and unless you have some massive changes, you can probably handle merge conflicts too.
Replies: >>226897
No you don't.  When you exit a program all the memory is automatically deallocated, that's how the Windows OS works.  Maybe you have to when you program in Linux, but that seems kind of insane.  As someone once said, the Operating System is the ultimate deallocator.  Once the program terminates, the memory is freed up for use.

I know in the documentation he says to do it, but it's basically just good "form" or "manners" as far as your typical video game is concerned, and it's a useful thing to remember if you want to do some other stuff before the program closes, in order to free up memory, or if you're running the program from some other program, in which case you might need to free up everything.
Replies: >>226908
Nope. Version control isn't smart enough to know what your vertex soup will actually look like when rendered so running merge algorithms intended for source code on it seems like a bad idea. As a workaround you can use gitattributes to flag blend files as binary so it doesn't even try to diff them.
Replies: >>226908
>just good "form" or "manners"
It actually depends on what kind of data you don't free. If it's something you allocate at the beginning of a level, and don't free when moving to the next one, you have a memory leak. Sure, if there's something you only allocate at the beginning of the execution and you need it during the whole lifetime of your game, sure, you can leak them (even though things like valgrind and such might bark at you), but to be honest that's a tiny portion of the data you handle. Everything else must be freed, or your game's memory usage will grow unbounded, until it either crashes or the computer becomes unusable due to excessive swapping.

Yeah, I didn't mean to manually merge two completely different meshes, but say you have one version where you changed one material a bit, and an other where you changed a completely different object's mesh, there's no reason why shouldn't you be able to merge them (you'd just need some proper text based format).
Replies: >>226933
Yes, obviously you free the memory you're going to use again during the lifetime of the program, but he said "before exiting the program".  If you're exiting the program, you don't need to do anything, because it'll all be deallocated automatically.  

As for memory management, you don't really need a pipeline for that, though there are memory management solutions you can use that make it easier (smart pointers, memory arenas, RAII).  I was specifically referring to, in the "form" or "manners" portion, how raysan in his examples that he provides, deallocates shaders and resources before exiting the program, even though it isn't actually necessary.
Replies: >>226935 >>227244
Here's the work on Teagan done so far. Give me a few pointers if you have any. 
Replies: >>227103
To be honest, that's exactly the place where you want to show how to properly free everything and not being a smartass and skipping things that are not necessary in your exact use-case, but necessary in 99.9% of the use-cases people reading your examples will have.
Replies: >>227085
My exact use case of exiting the fucking program?  Nigger what?  I'm sorry, does your game run indefinitely or something?  99.9999% of people will be closing your game at some point, which is *when* he is freeing the resources, not in the middle, not when the resources are needed for some other processes, but literally 2 lines before the return statement at the end of main.  Don't be a cunt trying to be smarter than you are.  This isn't an example of using that precise piece of freeing the assets, he does this in every single example program he makes, of which there are probably around 30-40.  Once again - this is good form.  It is literally unnecessary.  It is an exercise in remembering "hey, I allocated that memory, I should deallocate it now that I'm done with it", even though it will automatically be deallocated approximately 1ms later, when the program exits.  You know, the kind of thing an example/tutorial would do but is completely unnecessary for actual production code, and trying to do such will only make your life more painful.
Replies: >>227095 >>227097
Have you ever played a game that you had to restart every 1-2 hours because the developers were idiots who didn't free their resources and memory usage continuously increased while playing? But I guess you're also fine with chrome eating all the memory you have, since after you close it, it will be freed.
Replies: >>227103
>Arguing about inane shit like actually you don't need to free on exit
Just like make game you fags.
Are you illiterate, anon? Nothing you said is relevant to the discussion at hand. 
If you want to fight strawmen, at least make a game where you fight them instead of doing it here.

I'll probably give it a look later, but you'd most likely get a lot more feedback about the overall model if you at least posted some screenshots like before.
Replies: >>227111 >>227247
His problem is that the author frees the resources in an example where in that specific, simplistic case it's not 100% necessary, But an example should show how to use the api properly in a real-world case, it's not a code-golf challenge. If he never shows you how to free the shit, how would you do it when you actually need it (which is probably 99+% of real use-cases)?
And as I said, many programs leak resources like shit, you don't want to spread that mindset even further.
Replies: >>227129
But by showing it as "free stuff at the end of the program" you DO spread that mindset. Newfags see it and think that's the only time they need to do it when it's actually the one time they don't need to. Instead there should just be a separate example showing it as a step in something slightly more advanced that actually involves switching between resources. Actually show what you're trying to teach, instead of showing something entirely different and hoping a newbie will magically infer what it's actually intended for.
Replies: >>227143
>"free stuff at the end of the program" you DO spread that mindset
Doh. Okay, I have no cure about being terminally retarded, but if you have something like // free stuff when no longer needed before the free code, anyone with an IQ level higher than room temperature should be able to figure out that he needs to free that when not needed, and not just before the exit. If you move it to a separate example, 99% of newfags will miss it (unless they're veteran c coders) and will just think the engine/gc/whatever will take care of it. Because admit it, most normalfags these days learn coding in python, javashit, java, and other interpreted gc garbage, and not in c where your actions actually have consequences.
[Hide] (28KB, 749x449) Reverse
[Hide] (22.7KB, 749x449) Reverse
[Hide] (28.5KB, 749x449) Reverse
[Hide] (151.4KB, 1606x1091) Reverse
Spoiler File
(110.4KB, 1387x893) Reverse
Here's some 3 characters you'll meet in the next Nortubel update.
The art looks like dogshit but it's mostly because of a "character designer without actual art skills" way, the part where I trace over models in Blender and me in general trying to get shit done as fast as possible.
I do want to try to make them look less shit in cutscenes at least, if the overall "reference" art here looks a bit wonky (Specially Zarina because I want her to look good or show some intenions of that, specially compared to AAA games making characters look ugly on purpose).
And being the autistic pervert I am, I did a shitty, model traced art of some characters as an attempt to shill my game (Didn't work).
The thing is that I even have some "rules" for my characters due to a "fake crossover" nature as in "these characters are safer, these are mature, there are in a middle area etc", so this image technically breaks a rule.
But I can also claim it "non-canon" if for the fact it's not in the actual game.
I already added a "fanservice-y, even if still ugly" illustration in the game because it wasn't that explicit, so maybe the version with clothing could be added and make it more hidden.
Replies: >>227202
[Hide] (122.1KB, 415x497) Reverse
[Hide] (238.2KB, 789x296) Reverse
>I said "exiting the program"
Fuck I wanted to say in general.
I'd say an engine can read a file that describes a scene context with objects, then removes objects described from it from ram when another scene context is loaded.
Replies: >>227250
[Hide] (143.6KB, 589x620) Reverse
[Hide] (202.4KB, 783x749) Reverse
[Hide] (159.1KB, 746x757) Reverse
[Hide] (200.1KB, 751x743) Reverse
[Hide] (238.9KB, 803x737) Reverse
Replies: >>227355 >>227440
That's a better definition I suppose.  I'm not certain I 100% agree with it, but it's not exactly disagreeable either.  Also I saw all your deleted posts lmao.

Also, other guy, you're still retarded.  Once again, you free shit when it needs to be freed.  The very end of execution is not that time.  Don't know why you think you're gonna get a memory leak when you exit the program, but you won't.
Replies: >>227251
>when it needs to be freed
And when do you have that point in a simple example that only loads one thing and use it until it ends?
>you think you're gonna get a memory leak when you exit the program, but you won't.
I never said that. As long as you have something more modern than MS-DOS/16-bit Windows, the OS will clean up the shit you made in memory.
Replies: >>227541
[Hide] (1.2MB, 1873x980) Reverse
[Hide] (117.8KB, 1920x1080) Reverse
This week I got part of a Dragon's Crown Pro frame to be replayed on my PS4 emulator:
It currently outputs bits of the background. The garbage behind is the previous frame, which isn't being cleared for some reason.
The other parts aren't being drawn since they're in different command buffers, which the replayer isn't handling yet.
The game needed some buffer format runtime conversions, which were added to the shaders themselves. From further debugging, it will also need rectlists to draw rectangles correctly.
Performance sucks since the emulator is doing a render pass for each draw, along with copying textures between CPU and GPU each time.
Some other games' replays were tried but only We Are Doomed is (mostly) working.

>Frame dumper now exports JSON reports for each draw instead of each submitted command buffer
This solves an issue where it would run out of memory when handling games with more draws.
It's causing many files to be produced though, but that could be fixed by archiving everything into something like TAR.

Here's some frame stats:
>We Are Doomed: 5 draw calls, 3.1MB of frame data
>Hotline Miami: 83 draw calls, 11MB of frame data
>Dragon's Crown Pro: 518 draw calls over 4 command buffers, 92MB of frame data
>Gravity Rush Remastered: 4637 draw calls, 302MB of frame data
[Hide] (307.3KB, 480x571) Reverse
How the fuck did you model her hair?
I struggle with hair so much.
Replies: >>227380
[Hide] (6.5KB, 520x432) Reverse
>I struggle with hair so much.
I also struggle with hair a lot. I had to think a great deal about how I would do it, and while I am not 100% happy with the result I still think it looks really good. I used several references of her hair in fanart and I also downloaded some hair models from models resource to see how they do it. Specifically the model here https://www.models-resource.com/3ds/pokemonxy/model/18105/ really helped me a lot to model hairs. I just slowly but surely started with just the bangs. If you look here >>225198 you'll notice I don't show the top of the head, and that's because it wasn't there. I just slowly spun around until it was complete.

I had a hard time figuring out how I would model the top but I had a breakthrough when I realized I could model a ridge of thicker hair as you see in the first pic, and then off set it slightly to look wilder. I then filled in the remaining head with two half domes. Finally while playing killing floor I noticed the Cirno model was using some polygons in the shape of triangles to make spiky hair ends. So I tried that with my hair and it came out really nice. Making this hair was a journey to say the least. Pic related is how I did it. I just extruded a flat plane of one or two polygons and then joined the ends to make spiked hair coming off.
Replies: >>227440
Good job, anon. That's a hair any man would love to headpat..
Did you also sculpt on some of the spikes? They look like the have some mass to them here: >>227247
Replies: >>227447
Yes the bangs and the middle ridge of hair spikes are all thicker.
You never said it, yet you seem to be arguing that you need to free just before you exit the program.  You have implied, repeatedly, that my practice of not freeing assets 1ms before exiting the program will cause a memory leak.  I have never claimed you do not need to free assets at all, unlike how you think.  If you will perhaps manage to read, you can see that I say "freeing just before exiting is unnecessary", which you argue vehemently against because you're fucking stupid.

Once again, the tutorial was an example.  I am not arguing against the examples, they are good examples.  But they aren't an example of what you would do in production code.  They are self contained, with little idiosyncrasies like that which, in a real program, are completely unnecessary.  That which is unnecessary, is freeing right before exiting, not "not freeing at all" which you seem to think is what I'm advocating for because, once again, you can't fucking read.
Replies: >>227611
No, I'm still arguing about how you don't want to free shit in examples, where you're supposed to show how to use the API properly without doing dirty tricks. But then you somehow generalize from there to "you have to free everything just before exiting but not sooner" which is of course complete bullshit and I never said it.
>But they aren't an example of what you would do in production code.
Except they are. You want to free shit when it's no longer needed. That it coincides with the end of the program is irrelevant there. If you want to be lazy and skip 3 free function calls before the end of your program, do it, but everything else you should still free them properly.
Replies: >>227621
> If you want to be lazy and skip 3 free function calls before the end of your program, do it
Wow, it's almost like that's what I've been saying the whole time.  Glad you agree with me, retard.  Also, if you free everything you use before the end of the program, you'll have to free way more than 3 things.

It's really amazing how you can misconstrue "you don't need to free before you close" into "lol, this nigger doesn't free anything".  Are you being stupid on purpose or do you legitimately not understand?  Once again, the things in those example are being freed at the end of execution.  At the very end.  Not in the middle.  As in right before the program closes.  They are freed, in order to demonstrate.  Not because you need to, or because it's what you should do, but for demonstration purposes.  To clarify he is, in the example, freeing resources and closing out the OpenGL context.  You do not do this unless you are ending the process, unless in very niche circumstances.  This does not merely happen to coincide with the end of the program, you could not do any of these things except at the end of the program.


Note these two functions.  In a full on game, these functions might be necessary, for instance, if you are not going to use that texture or shader again (though typically you'll simply reuse the reference throughout the course of the game, with no need to free, as you will not be loading these textures or shaders again).  However, unloading textures and shaders is not necessarily something you'll be doing over the course of the game, especially ones you'll continuously reuse.  However, the use context for these functions is something you might possibly want to call over the course of the game.

This function call, however, is completely unnecessary.  This will never be called unless you are about to close the program, and raysan explicitly demonstrates how you should close out the program, and it isn't with CloseWindow (he uses a loop which only exits when an event is called that determines the window should be closed).  I would claim it would be bad practice to actually use this function at all, as you should be setting your game loop to exit rather than calling this function, as you might actually get unwanted and problematic functionality (for instance, calling it in the middle of serialization).

So no, these functions are not, typically (though they could be), called somewhere else in the code, they are explicitly called at the end of execution because it is the end of execution, in order to demonstrate what memory is being used, not because it is something you should do in code - it is for clarity.

Done arguing with you about this stupid fucking thing.  If you call these functions at the end of your code, you don't know how your code works and you just copy paste from the examples.  From an architecture point of view, attempting to keep track of every persistent thing that you usually won't need to free in order to free it at the end of execution is an exercise in futility and stupidity, and moreover, by attempting to track the lifetime of your memory by the start of the program to the end of the program, instead of explicitly within necessary lifetime, is a stupid fucking programming practice.  And to leave no room for ambiguity, a memory leak only occurs when you fail to free a chunk of memory that you then discard and fail to use again, most typically through some kind of function that continuously asks for new memory within the function.  Simply asking for memory that you continue to track and reuse does not constitute a memory leak.  The idea that every malloc needs to be followed, at some point, by a free is completely false - you only need to free the memory if you're not going to use it again during the lifetime of the program.  In the immortal words of a boomer, free ain't free (in cycles), so minimizing calls is a good programming practice, and you should hold a better understanding of how the memory works in your program than the simplistic and reductionist "malloc should be followed by free".
Replies: >>227647
>all this wall of text
Jesus man.
>could not do any of these things except at the end of the program
I don't know whether raylib supports anything besides opengl, but in directx/vulkan you can lose the device for many reasons (sometimes as simple as alt-tab) at which point you pretty much have to recreate the context (or just be lazy and crash, something which actually many games do). Sometimes you need to recreate everything if you want, for example change the renderer. Just because something is unnecessary for you doesn't mean there's no need for that.
In my engine tests, I create a completely new instance of my engine for each test, and tear it down at the end. Different tests are run with different configs and it's just easier this way to guarantee that there are no hidden dependencies between the tests.
>free ain't free (in cycles)
<not only microoptimizes, but then doubles down on it
Yeah, because that 0.1ms extra time in a code that will be normally executed only once counts. But if you're so anal about performance, you should be using custom allocators anyway, to decrease the number of expensive malloc/free calls (and possibly syscalls).
[Hide] (43.3KB, 885x601) Reverse
Working on Teagans little boots right now. The only daunting part of this will probably be making her fanny pack, and medallion.
[Hide] (1.8MB, 1920x1080) Reverse
Okay, with context it looks even worse.
Replies: >>227996 >>228073
It's just a door with the paint cracking from old age. It doesn't look that bad.
Replies: >>227998
I mean the shape of the door is barely visible, I guess the contrast between the painted and non-painted areas are too big. And I don't know what else, it just looks weird.
(The floor and the random untextured CC0 guy is just a placeholder.)
Also, do anyone have good resources of how old buildings look like? This specifically wants to be an (unspecified) old manor, but the place where the servants lived, so probably simple and cheap.. but I'm not a historian.
Replies: >>228360
[Hide] (582.5KB, 852x735) Reverse
Just found this channel with very impressive short 3d modeling streams and I wanted to share it.
He makes very stylized characters which you don't see often and look bretty good.
Replies: >>228021 >>228070
[Hide] (314.6KB, 1920x1200) Reverse
i recall some old history that he got banned from using zbrush somehow, not sure but yeah he has some good knowledge but you have to absorb it in his streams
Replies: >>228025 >>228028
>banned from using zbrush somehow
<click on a random video
<randomly seek into
<he rants about now only black people can say the n word, if you're a white you'll be cancelled
And this is just one random data point. I can imagine he did/said things people didn't like.
>he got banned from using zbrush
It's software. How do you get banned from using something you install on your own hardware?
Replies: >>228029
I'd like to introduce you to a company called microsoft
Replies: >>228062
Fixed some details, finished the boots, and added her cape. I just have a few more things to do before she's ready for texturing.
I'd like to introduce you to WindowsXPKg and windows genuine advantage remover.
>company's Maxon One subscription service
That's what you get for SAAS.
All the nigger has to do is get an older version that he can own.
His anatomy stuff is pretty good. I'm 2D only for now but it really helps with drawings.
[Hide] (382.7KB, 1125x1500) Reverse
[Hide] (37.9KB, 371x564) Reverse
I think cracks should be reserved for the panels in the middle. For everything else it's better to show faded paint on deteriorating wood.
Replies: >>228094 >>228135
[Hide] (945.4KB, 2316x1965) Reverse
Turns out I had picrel as a reference, but it doesn't even look like that. I guess I should look at references more often.
Anyway, any good setup for that? All these jewtube tutorials shill pureref, but I'm not going to install a proprietary electron garbage just to display a few jpg files. What I tried once is galapix as a way to get infinitely scalable image thumbnails thing, but all you can really do there is to point it to a directory and it will just display all the images, you can't move them around and group them and whatever. Maybe I should just make a big ass image in inkscape, but I'm not sure how well it handles many image files (especially the old version I have). Just don't tell me to use latex+tikz+large page size, and open the result in a pdf viewer, because I might do that

Also, why the fuck does your fucking word filter blocks webp images you stupid fish
Replies: >>228102
Krita and Blender have a reference image feature.
Replies: >>228104
This was made in material maker, so that doesn't solve my issue. And I'm not sure about krita, but blender's reference image support is only really useful if you want to trace over the images, if you just want to have a few reference photos for inspiration, it's not that useful putting them into your 3D view (they'll be more often in the way than being useful).
[Hide] (593KB, 446x877) Reverse
[Hide] (550.1KB, 444x881) Reverse
Dunno, maybe I should get rid of this voronoi altogether? Second image is just the AO used with some value noise to select between wood and paint.
[Hide] (632.8KB, 438x876) Reverse
I think this is the best I can. If it's still bad, I give up.
Replies: >>228171
[Hide] (77.2KB, 556x548) Reverse
I still don't see the door with the worn-out paintjob.
I think you should snap out of the procedural faggotry and just whip out a camera, take a photo then plug it into Photoshop just like any human being did before [CURRENT YEAR].
Replies: >>228172
>I still don't see the door with the worn-out paintjob.
Then look harder.
[Hide] (85.7KB, 609x729) Reverse
[Hide] (91.7KB, 697x734) Reverse
[Hide] (43.4KB, 246x737) Reverse
[Hide] (71.3KB, 760x740) Reverse
Alright TEAGAN is finished. I'm going to start texturing her tomorrow. Here's the completed model.
Replies: >>228176
Replies: >>228177
Teagan doesn't have massive beeg teetees like Alto does. Speaking of Alto I'm going to redo some of her mesh after I finish texturing Teags. I'm also going to add her crazy smile.
Replies: >>228180 >>228194
You should add modifier Decimate and lower the ratio to .005 instead so you can get some sweet UT PS2 models.
Replies: >>228181
[Hide] (64.4KB, 326x395) Reverse
>UT PS2 models
How did you know I was going to add Alto with her tits out in power armor for unreal tournament?
Replies: >>228182
Add her in SM64 and OOT too.
Replies: >>228184
Not before I add them into SRB2 Kart.
[Hide] (1.6MB, 1419x2048) Reverse
If anything they could be smaller. I never really pictured teags having more than "realistic" large breasts. Muscular girls usually have smaller breasts when the artist isn't going full amazon mode.
[Hide] (1.7MB, 1920x1080) Reverse
I had a weird idea and added some geometry nodes to randomly move vertices a bit. Like a displacement shader, except it works for eevee too. (What is so hard in making a vertex shader next to the fragment shader is beyond me. Or at least a simple parallax mapping)
I guess I'll label this as complete. At least for now.
Also changed the door to be a bit more yellow, so it doesn't have exactly the same color as the wall, but actually I'm not sure which one should be more yellow in real life.
Replies: >>228319
That looks really good anon.
Bump my question from here.
I've been randomly reading the books linked from here
they contain a lot of info, but unfortunately very few on the actual topic I'm interested in (or I just don't find them in this digitized mess).
It still beats me how could the british build such a huge empire when they can't even count/measure things
After replaying a bit of Copy Kitty and remembering how flawed it was despite the obvious passion the devs had, I remembered that I'd had the idea to make a game inspired by its mechanics.  The premise of Copy Kitty is similar to Kirby 64, in that you can take powers from enemies (which drop when they are killed) and combine them to form new powers and weapons.  Kirby 64 has 7 base abilities and 28 combinations, plus a unique shmup weapon during the final boss.  Copy Kitty has like 20 base shot types and more than 160 combinations, plus many unique weapons that can't be combined or which are only used in boss fights.  Obviously not all these combinations are created equal; some are annoying and weak while some fill the screen with tornado giant dragons that shoot waves of bullets.

CK has a lot of fat to be trimmed, and the platforming and character control are wonky (your basic shooting action has recovery on the end, you have a landing animation if you fall too far, you have an invincible Dark Souls dodge roll, you have a kick that can destroy/reflect incoming shots).  The game also suffers from feature bloat and being insanely hard to parse when stuff is happening onscreen.

Combining abilities is fun.  It makes you feel like you're exploring the game's mechanics as much as you're exploring the world.  I want to take that fun feeling and make something more polished with a greater focus on the behavior of each individual ability and ability combination.

Magic Shotgun
I brainstormed about this a bit and came up with the idea of a magic double-barrelled shotgun which can take different types of magic spells as ammunition.  You can hold two types of shots, and fire each one individually or in combination.  With nothing in the chamber, you get a basic muzzle flash blast, sort of like a melee attack.  The movement should be pretty simple since having 3 shoot buttons and a jump button will eat up most of the input complexity.  A standard jump, a low slide that can dodge standard bullets, and perhaps some kind of wall jump.  No double jump since I want the shot types to have some kind of movement effects.  Progression will be level-based with no open world or backtracking/exploration, since that would get way too complicated.

As for the shot types, here's what I have off the top of my head:
- Spread: Fires a spread of bullets outward from the point of origin.
- Snipe: Fires a single long fast/hitscan shot that pierces through enemies.
- Shockwave: Fires an energy wave along the ground that follows the terrain.  Can "jump" off of slopes.
- Rocket: Creates a high-damage projectile that hovers in place for a second, then fires forward quickly.
- Seeker: Fires a continuous stream of low-damage flechettes that home in on nearby enemies.
- Bomb: Lobs a bomb upwards, which travels slowly and explodes on contact.

According to a quick online calculator I found, If I allow double stacks that would be 6 base weapons + 21 combinations (or 15 combinations if double stacking is not allowed).  This seems like a lot but I think it could be managed well enough.  I want some combinations to be useful for things other than damage, so if you combine Rocket and Snipe then you would get a "rocket takeoff" attack where you create a huge explosion beneath you and then go careening upwards.  Snipe + Bomb would create an enormous explosion at the point of impact, but no longer pierce enemies and have so much kick that it sends you in the other direction.  Bomb + Rocket creates a giant bullet that travels very slowly but can be used as a platform while it's moving.

Other combinations would be more for just shooting stuff.
Bomb + Spread would throw a fan of shurikens/knives that detonate after a short delay.
Rocket + Shockwave would create a rolling explosive shot that stays inert for a second but then accelerates quickly.
Rocket + Bomb would toss up a large bullet that carpet bombs the area in front of you.

So, yeah.  I'm still in the idea guy phase but if there are any points of reference or input you can give then I'm all ears.
Replies: >>228445
[Hide] (368.8KB, 1920x1080) Reverse
>magic shotgun
Or you could do what they did in Stranger's Wrath.
Replies: >>228451
I had forgotten about Stranger's Wrath but it's a good game and I love the combination of melee combat and FPS stuff.  The only issue is that the player has so much freedom with double ammo types that you can't really build challenges around any particular one past a certain point.  People can just choose the best combination and switch at any time, which isn't the kind of gameplay I want.
Replies: >>228453
>I had forgotten about Stranger's Wrath
Everyone has.
[Hide] (195.2KB, 657x623) Reverse
My lack of ability to make decent textures is making this look like crap.
Looks fine, just need some little fixes.
Does Teagan have FAS?
needs a nose job, use a 3 point lighting (pretend you're in a photography room)
show off more of the eyebrows to get a more feminine look, because this just looks like a dude
I got my little fun game to compile. I wrote it in the highest, most advanced programming language I am somewhat familiar with, basic. But if I tick the option "include debugging code", the compiler gives me a memory overflow error. I think I may have to clean up the code, make the main module smaller or something. It's a bit on the spaghetti side, and variable introductions aren't very professionally done. But it compiled, so I'm happy.
I've got a few questions:
Is "LOOP UNTIL TIMER > start + .01" good enough way to slow down loops for animations, movements and such? Do all loops need a slowdown? Should I bite the bullet and learn something advanced like Pascal? What kind of license, if any, should I include in code/readme?
Replies: >>228543
Learn FORTH.
[Hide] (320.3KB, 437x691) Reverse
[Hide] (239.8KB, 1920x1080) Reverse
Looks too flat and the eyes are a little too far up.
Add some shadows to make the skin texture to make it more 3D or make them hard shadows if you want a more cel shaded look.
[Hide] (87.5KB, 360x477) Reverse
I broke down and completely redid Teagan's face because I just wasn't happy with. I'm still not completely happy with it, but it's leagues better than it was.
Replies: >>228839
Something's wrong with the proportions of the face, or the shape of the head, I can't tell, but there's definitely something off about it. The eyes give the impression of being asymmetric. The nose and chin shadows are too pronounced.
Speaking of blender, is there any good source for getting models, or will I have to model everything on my own?
Replies: >>228862
Have you tried models resource? They have a decent selection.
Replies: >>228875
Hmm, interesting, but it seems to focus on player characters, while I'm more interested in props now. (Also they're generally rips from very old games, a bit better quality/resolution would be better).
[Hide] (141.4KB, 1319x872) Reverse
>make a 2.4m tall room
>make a 2m tall door
>make a 0.5m tall window
>put the window over the door
>be surprised that it doesn't fit
This is getting ridiculous. Should I walk around with a tape measure and measure everything I see and hope I'll get a feeling for sizes somehow?

Anyway, any good glass shader for windows? Searxing it I found either sites with buy this shader pack for only this many shekels goyim pages, or 15 minute long jewtube videos where he uses a single principled bsdf node and nothing else in a way that doesn't work in eevee, and nothing in between.
Anyway, I've dinged together a somewhat believable shader with a mix of Transparent BSDF and Principled BSDF using some fresnel shit, since transmission doesn't seem to work properly in eevee. But it kinda works, so probably I shouldn't fuss too much over it.
Replies: >>228930
>Anyway, any good glass shader for windows
Glass had its own shader, both for eevee and cycles. Also you can just use principled with reduced alpha, no one will notice. Or more proper low roughness + transmission, should give the same result. You should also change IOR, but honestly, only pedantic assholes can notice it. Its not like eevee can simulate glass properly and neither is cycles, even if they brag about new features. 
Also you can use your scratches as normal map, via "bump" node, I think it works better than "normal map" node. And you can adjust depth for it, Its most likely will look better. Also in textures, usually you want "cubic" or "smart" texture interpolation, it really solves a lot of visual bugs. 
>metallic 1 glass
Its opposite of glass. 
>This is getting ridiculous. Should I walk around with a tape measure and measure everything I see and hope I'll get a feeling for sizes somehow?
Just use realistic sizes, its important for rendering too, and you can see them in object properties.
Replies: >>228932
>Glass had its own shader
I tried that first, but it was even worse. I think I read somewhere that it's not supposed to work well in eevee anyway.
I don't want much, just something that's mostly transparent, but still reflects some light, and add some dirt/scratch overlay.
>Bump node
Isn't that the same as normal in eevee? And even in cycles you have to subdivide your mesh as shit because it only ever transforms vertices.
>Its opposite of glass. 
But if I set it to 0, I get no reflections.
>Just use realistic sizes
Yeah, but my problem is that I don't know what is a realistic size in the first place.
Replies: >>229006
[Hide] (139.6KB, 214x452) Reverse
[Hide] (59KB, 794x770) Reverse
Took a whole day just for this shit.
I don't think 120px is tall enough, how the fuck does GGXX look good.
[Hide] (10.5KB, 276x68) Reverse
>I don't think 120px is tall enough
For the love of God, STOP, 120px is insanely big for pixel art, you can easily achieve what you want with 38px.
was going to say this, that's way too tall
[Hide] (18.7KB, 431x256) Reverse
>120px is insanely big for pixel art
Not if you're making a full-screen CG.
Replies: >>228953
learn basic 3d, then rotoscope on top of it just like sf3, later kof games, etc. or even better just cheat and make it into pixel art like dead cells
Replies: >>228954 >>228956
[Hide] (2.9MB, 320x240, 01:59)
[Hide] (2.7MB, 512x320, 01:59)
[Hide] (435.2KB, 540x384) Reverse
[Hide] (474.3KB, 540x385) Reverse
[Hide] (1MB, 721x504) Reverse
in a static scene the sky is the limit, you are free to do whatever you want, but, if you want fluid animation, drop to 32px-38px, 48px is the limit for a normal character.
Replies: >>228954
[Hide] (17.7KB, 276x68) Reverse
[Hide] (45.6KB, 142x270) Reverse
But then I can't have the face be well defined at those proportions. I want to be able to do eye animations and that's already hard at 120px. 
Even if I was good enough to make sprites like that that doesn't really work for the proportions I'm going for unless I get rid of the face or have the eye just be 1 red pixel. 
My autism prevents me from rotoscoping. 
Guilty gear looks good at a hundred something pixels. Obviously it takes more work but it's doable in a fast paced game.
Replies: >>228955 >>228958
>My autism prevents me from rotoscoping. 
I don't see anything wrong with rotoscoping, previous GG games & other ports have shrunk down the size of the sprites and it still looks recognizable. Overall you just have to count how many frames one character could have, along with the enemies you'll have to also animate as well.
Replies: >>229003
[Hide] (17.1KB, 480x691) Reverse
[Hide] (43.8KB, 480x691) Reverse
[Hide] (17KB, 480x318) Reverse
[Hide] (13.1KB, 480x318) Reverse
[Hide] (26.6KB, 480x460) Reverse
Replies: >>229003
[Hide] (8.1MB, 2212x3141) Reverse
[Hide] (14.7MB, 2226x3132) Reverse
>Guilty gear looks good at a hundred something pixels
The GIF of Axl you posted is the result of years of artistic effort, refinement through a couple art styles at different resolutions (GG, GGX, GGXX), and a very strong sense of identity.  Daisuke was a true renaissance man in his early years.

Even then, there are GGXX sprites that look weird or deformed. I'm sure it took the artists thousands of drawings to settle on something good to turn into a sprite.
Replies: >>229003 >>229014
[Hide] (1.9MB, 1920x1080) Reverse
[Hide] (2.2MB, 1920x1080) Reverse
Those weird shadows next to the doors start to bug me. Actually, why are they so big? If the lights were just next to the wall, I'd understand, but they're not so close. I guess I could disable shadows for the frame of the door, but then it wouldn't have shadows on the door either, so I should only disable the frame casting shadow on the wall, but I don't think eevee can do that.
But then, the same thing happens in cycles, even if it's a bit less noticeable there (after I set the lights' power to half, why are lighting so fucking inconsistent between eevee a cycles). Sigh, lighting was a mistake.
Replies: >>228974 >>229001
The doors look really nice, anon.
[Hide] (1.2MB, 1920x1080) Reverse
[Hide] (123.8KB, 1920x1080) Reverse
[Hide] (155.8KB, 1920x1080) Reverse
Been progressing a little with my PS4 emulator:
Rendering of Dragon's Crown Pro frame replay has improved (along with We Are Doomed's). It's still only doing the frame's background though.
Tried Sonic Mania as well since some other emulators feature it. Seems like rendering is done on CPU, using the GPU just to blit and upscale the resulting image. However it's not emulating properly yet.
[Hide] (1.2MB, 1280x720) Reverse
[Hide] (1.2MB, 1280x720) Reverse
[Hide] (1021.2KB, 1280x720) Reverse
[Hide] (1MB, 1280x720) Reverse
Hello anons, I was working on my farming game for a while. I am trying to currently implement muh immersive way of processing materials for ores but I don't know yet how I should animate this whole thing. I can animate part of it in blender that's not the problem but I am not sure how I should animate the rest like particle effects getting spawned and being timed with the animated model, so I didn't get around adding those yet.

I'm also using also substance painter and krita for texturing, with the final plan being to use stylized textures that I made in krita. I didn't added that many items yet since I was busy getting the code to work first. I also need to figure out how to make a proper grid system for building props. I have a grid system already in place but it's not working correctly at the moment like it is janky.

Though for now I will have to pause my farming game since I was supposed to work on a godot FPS project instead with a friend of mine. I got distracted with my farming game because I didn't wanted to pay up 6 million shekels for gmod mods just so that I can earn money from smelting ores and oil refining and what else there is.

thanks for joining my TED talk.
Replies: >>228984 >>228989
textures look rather grainy for the terrain
Replies: >>228986
It's just a placeholder map for development purpose.
>I am not sure how I should animate the rest like particle effects getting spawned and being timed with the animated model, so I didn't get around adding those yet
The way I've seen a lot of 3D games handle this is just tying events to the animation cycle, so when you reach a certain thing happening (like a piston making contact with something) then you play the particle effect from the location of the piston.
Replies: >>229004 >>229016
This is a message for the anon who was testing how my models animated and gave me some advice. For some reason I can't seem to get the automatic weights option to work despite watching a few tutorials all of which were completely different despite being 2.79 of course. How were you able to get it to work?
Replies: >>229001
>shadows are too dark
You can use bigger light source, and give renders more bounces. I think "light clumping" might help with render time/result. Alternatively just use "fake" light source where you want things to be brighter. 
Just model around rigify rig, or adjust it afterwards. Make a model> add rigify rig> scale it for proportions> apply scale for both the model and the rig (very important)> edit the rig to match the model as well as you can> press "generate rig" button in rigify bones, it will make you a proper rig> parent the model to a rig with automatic weights, and you can delete the original skeleton.
Replies: >>229002 >>230515
[Hide] (1.9MB, 1920x1080) Reverse
>give renders more bounces
But this is eevee, it doesn't really do bounces. Maybe I should try lightmap baking.
>bigger light source
I've increased their radius to 0.5m, now the shadows look much better... on the other hand look what happened to the reflection in the window. AAAAA!
Also I should do something with the stars in the window. Right now it's just a plane with the stars texture behind, which is fine if you have a single shot, but if you move around it's weird that the stars doesn't move. And HDRI in eevee affects everything, even things that are supposed to be in shadow. (Plus I'm not sure using a full blown HDRI just for that tiny window is a good idea)
[Hide] (54.7KB, 1024x600) Reverse
[Hide] (4.7KB, 125x157) Reverse
I tried redoing the sprite using the blob method rather than just tracing the outline. This time it was faster but I'm still not feeling it, it just doesn't look quite right. 
I guess the bigger problem is that I don't really know what I'm going for with the visuals. Maybe I should just drop the animations for now and work on concept art to figure out what I'm trying to do. 
By autism I mean autism. I just don't want to do it for no particular reason. 
That's the downside of working with physical medium. As a digital artist I can abuse the shit out of the select tool and invert the canvas to prevent too retarded proportions.
Replies: >>229010
So basically I need to call functions from the animation somehow in order to spawn particle effects then? ogg I will look into that.
Replies: >>229010
>Its opposite of glass.
>But if I set it to 0, I get no reflections.
Scratch that, I've set it back to 0 and still have reflections (even if with less intensity.)
Is there a good, not gimped description of all the parameters what this principled bsdf/pbr shit have and what they mean? Blenders own documentation is pretty terse and I'd rather not watch 30 minutes long jewtube videos just to learn that "base color is the base color, and metallic is the amount of metalness lel".
Replies: >>229011
>abuse the shit out of the select tool and invert the canvas
Please don't tell me you're still working in Paint.

Not necessarily "from the animation" but there should be some way of knowing what part of the animation is playing, whether the animation is playing or not, and its speed.  I think a lot of older games had independent timers that would simply call sprites and put them into memory.
Replies: >>229014 >>229016
>metallic is amount of metalness
And its binary in actuality, you are not supposed to put it in any other position, but you could. But the only option is to watch a few videos, each a few hours long, half of them wrong on many points. There are a hundred nodes, each with a lot of parameters, and quarter of them work differently in different engines. You just learn as you go. But most nodes are for very specific purpose, so its not that important to know them. Bump, roughness, color are the most important nodes, sometimes you use metallic, and very rarely sss, gloss and transmission.
Replies: >>229036
>Please don't tell me you're still working in Paint.
No, I meant in general. With digital you can flip the canvas to catch fuck ups like that big head in >>228958
and you can use the select tool to shrink that part and make it fit.
[Hide] (194KB, 1280x855) Reverse
Ehhh I am gonna ask on Godot shitcord for help and see if they know how the hell I am supposed to animate this whole thing, since I want to know it anyway how to add fake animated fluid like in real life smelters for example and rocks being thrown around.
[Hide] (325.6KB, 900x2418) Reverse
>watch a few videos, each a few hours long, half of them wrong on many points
Nah, I think picrel is what I should have read before.
Replies: >>229104
[Hide] (92.9KB, 259x195) Reverse
>generic description of pbr
Its more than a decade outdated and not focused on blender at all.
Replies: >>229107
>more than a decade outdated
Then show me something better. Sure it doesn't go into anisotropy, sheen, clearcoat, transmission, but still better than what most blender artist soyboys out there can produce. I'm not interested in hours long jewtube videos with 1/10th content.
>not focused on blender
But that's supposed to be one of the goals of this whole PBR shit, that it's the same everywhere where PBR is supported. Sure, there are going to be differences between implementations, but the basic should be the same everywhere.
Replies: >>229109
That paper can be summarized in 5 minute explanation, it doesnt really provide anything useful, unless you cant understand what a roughness and normal maps are. 
>same everywhere
While it supposed to be the same, regular 3d engines and raytraced engines are extremely different, and while same principles apply, virtually identically, the results are very different.
>the basic should be the same everywhere
But these basics are extremely basic, and they no longer need any real explanation, everyone does it already. It is an interesting topic, but its more of historical value these days. There is nothing wrong with reading it, but if you want to learn some practical skills, jewtube videos seem to be the only way. Just find the least annoying ones to watch, and even if you know what nodes do, a lot of it relies on tricks, to get results you want. 
And btw, most properties on principled shader are useless 99% of the time. Specular value is always 0.5, you supposed to adjust ior to get proper realistic values. SSS is only useful for noses, ears and hands, in virtually any other situation its just a waste of rendering time. Transmission is only useful for glass objects.
[Hide] (41.6KB, 1024x600) Reverse
Is this scale too small? I kind of forgot that pixel art needs to fit the screen to not look like shit and this is the scale I'm stuck with if I don't want an even taller sprite. 
Here's a demo with a fucky camera. ' to change the zoom between the old scale and the new scale. 
win https://litter.catbox.moe/zyx1qb.zip
linux https://litter.catbox.moe/qgqh74.zip
Maybe I should just put aside my hatred for non-pixel art 2d platfomers and drop the pixel look.
Replies: >>229327
I think it looks fine based on the screenshot.
>hatred of non-pixel 2D platformers
You should only hate them if they look bad.  There are games like Braid that manage to combine good 2D gameplay with a more painterly/illustration graphical style.
Replies: >>229330
The only reason I hate them is because I played Braid once and thought it looked so upsettingly ugly it overrides any good looking game I know when I try to think up non-pixel 2d.
Replies: >>229336
Heh.  I guess Braid's style is very stiff, and combined with the time mechanics it can look pretty cheap and ugly sometimes.  But my overall memory of the game is that it looks quite good for a game from 2011/2012.

2D games that don't use sprites or 3D models but still look good are pretty rare.  In many cases you get something with a lot of awkward paper doll animation or a jarring mix of 2D running around in 3D like Shantae Half-Genie Hero.
Replies: >>229432
[Hide] (198.8KB, 388x630) Reverse
[Hide] (26.8KB, 94x145) Reverse
[Hide] (25.2KB, 91x162) Reverse
[Hide] (9.2KB, 114x188) Reverse
[Hide] (6.7KB, 134x144) Reverse
>Braid's style is very stiff
Seeing it now it does look good when it's still but the animations still look wrong. I don't know what it is about it that upsets me so, maybe it's just the character proportions. 
>2D games that don't use sprites or 3D models but still look good are pretty rare
The only ones I know of are skullgirls and cuphead. I tried looking for more games but the results get clogged with 3D games with 2D cameras when you exclude pixel art. 
I think I might just go for the hueg pixel art. I don't think I'm skilled enough to make non-pixel sprites look as good, and there's a precedent for big girls in indie.
[Hide] (180.8KB, 256x256) Reverse
I really should stop procrastinating. But is it really procrastination if my head is killing me?
Replies: >>229463 >>230096
[Hide] (206.9KB, 518x575) Reverse
>made a 8192x8192 texture for the floor of a small room
I might be going overboard, I think.
Replies: >>229940
peak tism
Replies: >>229944
Yeah, my fear is what happens if someone has a 8K monitor, despite the fact I personally only have a fullHD one.
Replies: >>229950 >>230070
Make all your textures in .svg format and have the game rasterize them on load to match the player's resolution.
Replies: >>229951 >>229965
And while I'm at it, also render the game to svg, so screenshots are infinitely resizeable, right?
Replies: >>229965
That is actually a good idea. You not only get to support all resolutions in existence, but also get reduced file sizes as SVG is a pretty small format. The downside is you can't stimulate fine gradients and complex textures without mixing in some procgen.

You jest, but there are plenty of games that save screenshots/replays as "game states" which can be rendered and stored at any resolution.
Replies: >>229967
>mixing in some procgen
Yeah, SVG itself is pretty crap if you're going for anything but the most cartoonish look. And this texture is procgen, but it took like a minute to export to an image because during the process it ran out of VRAM and started copying textures back and forth between the system ram and VRAM... But I guess if someone has a 8K monitor, he also has money to buy a GPU with 100+GB RAM, right?!
Replies: >>230034
You should shrink the generated texture image as much as possible and let the renderer do its texture filtering magic, you'd be surprised how perfectly acceptable a filtered texture looks.
Also, an 8K screen does not mean 8K textures. Aside from very few exceptions, your texture size should be a quarter of the screen resolution, if not less.
Replies: >>230042
But I can see the difference between a 4K and a 8K texture on a 8K render. Actually, even with the 8K texture, since you're not seeing the whole texture at once, you're still zooming in.
Replies: >>230070
>8K monitor
The most ludicrously overpriced monitors on the market are only 4K.  8K is a fever dream being sold to idiots with too much money by companies who don't know how to make good products.
You aren't seriously getting worked up over this, are you?  Do something more productive or meaningful.

>But I can see the difference between a 4K and a 8K texture on a 8K render
I can guarantee you that nobody who doesn't have acute autism will care about texture quality in screenshots to this degree.
Replies: >>230076 >>230093
I have to wonder if its worth my time to use index+palette format to save space. And whenever 2k atlas is good and universal enough, I could not find any data on "minimum maximum" texture size of hardware still in use. So stop having fun, this is getting annoying. 
Some even sell 48bit color monitors.
Replies: >>230093
[Hide] (207.4KB, 321x932) Reverse
[Hide] (198.9KB, 495x925) Reverse
>nobody who doesn't have acute autism
Then what about me?
In somewhat unrelated news, I made a G9 LED light bulb that's ridiculously detailed for something that will likely never take up more than 10 pixels on the screen. I also downloaded this fucking IEC60061  standard, so the G9 connector has correct sizes. But hey, combined with ridiculous amounts of bloom, it kinda looks like a light.
I don't know how, but someone please stop me.
Also, is there any use in this model, or should I throw it out?

>index+palette format to save space
It's a PITA on the GPUs made these days, but can be useful if you want palette recolor effects. If you're doing some pixel art game, you're not going to save any meaningful amount of memory by compressing your 32x32 pixel sprites, and if you want to convert normal textures to indexed, they'll likely come out horrible (though obviously check out pngquant, it produces much better results than gimp and other naive RGB->indexed conversion algorithms). At that point you're probably better looking into S3TC compression (which needs EXT_texture_compression_s3tc extension in OpenGL, but supported by virtually every GPU) (there are newer variants, but they're not as widely supported).

>2k atlas is good and universal enough
OpenGL until 4.0 only requires 1k textures, opengl 4.1 bumped up the requirement to 16k. But realistically, there's this thing that collects opengl info from shitloads of users, and the only renderer that only supports 1k textures are the built-in software renderer in windows vista (that by the way only supports opengl 1.1):
Realistically, anything that supports opengl 3.3 should also support 4k textures, and anything that supports opengl 4 should support 16k. (But at that point you're probably just wasting memory).
Replies: >>230095 >>230096
[Hide] (11.7KB, 300x94) Reverse
>Then what about me?
The only one who can help you is yourself.  Please get a grip.
>ridiculously detailed for something that won't take up more than 10 pixels
This is a waste of resources and time.  Having made the model, there is no reason you shouldn't keep it since you may be able to reuse it for something else or use it as reference, but you ought to consider the practical output of your workflow.  If you make everything to satisfy a hypothetical 8K monitor with some insane graphics settings then you'll never finish anything because by the time you do there'll be another graphical benchmark to reach.  Achilles and the tortoise for 16K resolutions on 500Hz refresh rate monitors.
Replies: >>230108
>It's a PITA on the G