TIL that animating a color with with property [hdr] in unity records it with .rgba bindings. Problem is hdr colors are linear while .rgba bindings are srgb. So you won't end up with the color you picked. #unity
github.com/d4rkc0d3r/Un...
use d4rk's Avatar Optimizer. You simply add it and click a button. Will greatly clean up your project (reduce meshes, materials, blend shapes, and so on).
github.com/d4rkc0d3r/d4...
I've also added a little button to promote my ko-fi on there now. It's been long overdue that I mention that somewhere more prominent than just linking it in my github profile.
Anyways, get the new version here: github.com/d4rkc0d3r/d4...
When using shader toggles it now adds material swaps to a disabled material to each slot that comes from only one source mesh. This way materials that can't get merged can get their draw calls removed when their source mesh would be disabled.
I'm finally done with version 4.0.0 of my avatar optimizer.
With this version I am dropping support for older unity & vrc sdks completely. Make sure to update your projects!
The main feature is a whole new UI to debug why it can't merge material slots for the advanced users out there.
An update on AppleGI:
@appleblossomvr.bsky.social
and special thanks to @rainy.fans for modeling
I just pushed out v7.0.0 of my shaders, and while I wanted to do a full on fancy trailer and a bunch of other promotion stuff - I didnt want to delay it any further. So it is live now! 350+ commits, hundreds of lines of changelog, all free
shaders.orels.sh
Happy Holidays! ❤️
#unity3d #shaders
My "No Graphics API" blog post is live! Please repost :)
www.sebastianaaltonen.com/blog/no-grap...
I spend 1.5 years doing this. Full rewrite last summer and another partial rewrite last month. As Hemingway said: "First draft of everything is always shit".
32bit IMAD on nvidia is same max throughput as shift or bit wise ops since turing. @gegellibu.bsky.social specifically asked me for the nvidia throughput table before posting this.
docs.nvidia.com/cuda/cuda-c-...
Taking another look at the morton order code @vassvik.bsky.social posted, one can further reduce the amount of instructions required, using one of my favorite bit fudging techniques: Bit cloning using multiplication!
It sometimes lets you replace code like x | (x << a) | (x << b) if a, b are fixed.
I want to put a neat morton trick I have below here too. A way to get the next morton index without converting to 1d and back.
Some random Morton Code tricks:
If you've worked with Morton codes you've probably seen something like the this GLSL code to decode a Morton code, which end up at roughly 27 integer operations to generate the equivalent 2D position.
Code transcribed from fgiesen.wordpress.com/2009/12/13/d...
Decided to go back to making Shadertoy fluid sims for for a bit.
This time I improved my particle clustering cellular automaton SPH fluid sim to support multiple species particle merging and splitting. Also some fine tuning of the physics.
shadertoy.com/view/tfyyWD
Announcing: New @steampowered.com Hardware, coming in 2026:
Steam Controller
Steam Machine
Steam Frame.
Watch our jazzy announcement video and wishlist now: steampowered.com/hardware
I don't use twitter anymore.
Seems like my dms here were set to only followers by mistake so I fixed that.
Actually IIRC constant buffer can be pretty slow when doing non uniform access on it. So maybe thats it?
That is very surprising to me. When doing hlsl with fxc the non const version would be pathological slow since it compiles it as pushing all values into indexable registers just to read the one value out.
While const version just gets put into immediate constant buffer.
Had this knocking around in my head for about 6 months and finally got around to writing it out and filming it. This is my first time making a video essay and I haven't written VO since college but I'm happy with how it came out. Let me know what you think!
Filmed in #vrchat
Edited with #blender3d
I had something similar in my repo for a while. Completely jumbled up release order. However it was the correct order when I looked at the list with an incognito window which is really weird.
what if your avatar's minimum brightness was set to zero, it'd be pretty cool
Been playing with some wetness effects for a #VRChat world recently. Took the math of @mochie.bsky.social 's water gerstner waves, rendered it to a texture, and then did some simple decay on it via Graphics.Blit. Its a little jank but I've stared at it long enough that I think its working #unity3d
Man you project is cooked or something. I've never experienced or heard of anyone else having those two issues before.
Especially the play mode one. I don't think I ever upload without going into play mode beforehand to test.
Light Volumes v.2.0.0 released!
Now it has Light Volume based super optimized analytical Point Lights, Spot Lights and Area Lights. (Up to 128 at the same time)
Ability to project images and cubemaps.
Also baked shadow masks and even more features!
github.com/REDSIM/VRCLi...
The hair has to make way for the cuddles.
I made a video about my (extremely crude) software rasterizer. You can watch it over here! youtu.be/yyJ-hdISgnw?...
You basically use the samplers and tex2dlod anytime that is defined and texture2d and .Load whenever it isn't defined
blegh, personally I would simply not support surface shaders. But if you really want it you can do it by hiding it from the surface analysis thingy with quite a bit of macro spam by checking SHADER_TARGET_SURFACE_ANALYSIS
I took a quick look at it. You used 3 sampler2D in a linear nx1 way. Using just texture objects and .Load would probably be better. 3 samplers add to registers which means you might get knocked down in occupancy while just doing textures would not. Also 2D textures might not like a nx1 layout at all
Technologies change, the problems stay the same. Was trying to improve the results with @sim.red Light Volumes
Original clip by @crowscrowscrows.com #unity3d
