Randy Gaul

@svadrb @maciejhalber I just asked Claude to make the app with fixed function OpenGL. Also gave specifics about the data to render and collect. I was searching for an edge-edge separation function that would be easy to root find over time

@RandyPGaul @maciejhalber What kind of process do you use to create something like this with Claude? I haven't had much luck (or skill, I guess) myself.

cc really amazes me each time I have make one of these one-off visualization tools

@maciejhalber This is just pure asking claude to use fixed-function OpenGL to code something on-the-spot! Nothing fancy at all!

@RandyPGaul wow, this is awesome! is this your gui framework based on cute libraries or something else?

Cylinders🤦

Finally, the very heavy big ball sim is working.

@fdadwd I was waiting for someone to say that

@RandyPGaul AI for reading paper and long blogs is great.

AI assisted coding — if you aren’t using AI for software engineering then you’re being swallowed by a whale. You don’t want to be swallowed by a whale. randygaul.github.io/coding/2026/02…

@konigssohne I’ll give that a try 😊

@RandyPGaul One thing you could~ add is chromatic aberration. It would make the crystal light a little more realistic.

New lighting system is going pretty well - using Holographic Radiance Cascades (arxiv.org/pdf/2505.02041)

@kodingnights I’m just running it in real time for my game, it solves all the lighting within the HRC grid one-shot so it’s suitable for fast paced games too. For larger res games you can do tricks like rotate which quadrant you update, upscale the lighting. HRC is a fast algorithm

@RandyPGaul Very pretty but I guess some kind of post processing?

@dreambuffer Claude

@RandyPGaul Amazing, how did you implement it?

@kodingnights Typically you’d only render lower resolution and upscale. Vid I posted is doing 480x270 compute at 300fps

GPU time implication, from the paper: Full HRC, RTX 3080 Laptop): 2048² -> 33.9 ms. - Extrapolated to 4096²: about 136-148 ms per frame. - That is not realtime for a 60 FPS budget if you run full-map HRC every frame. VRAM implication Using paper memory formulas with quadrant memory reuse: - 1024²: ~168-224 MiB - 2048²: ~720-960 MiB - 4096²: ~3.0-4.1 GiB Quite unusable in a game, in other words. My own implementation has a cost of 7 ms per frame at 4096, however it is not full HRC and not as pretty.

WIP serialization + versioning API in C, lmk if you have thoughts/feedback! gist.github.com/RandyGaul/b65c…

NEWS! raylib custom Software Renderer backend [rlsw] has been finally merged! 🚀 For the first time in this 12-years journey, there is a NO-DEPENDENCIES path available for raylib rendering, no platform library (direct Win32 implementation) and no OpenGL required! 🤯

@raysan5 That’s so cool! It’s truly cross platform then. Is it possible to make a “shader debugger” by stepping through CPU function that’s functionally equivalent to shader functions?

✅ Design a 2D renderer from first principles ✅ Break all the rules (in a cute way) ✅ Getting swallowed by whales A Tour of Cute Framework's Renderer randygaul.github.io/cute_framework…

大好評発売中の爆弾いもむしの探索アクションパズル、「Öoo」が20％セール中です！！！ 1日で遊びきれる短編作品でありますが、中身はぎゅっとつまった濃密な体験になっています。 ギフトにもぴったりです！ 各種お得なバンドルもありますよ👇

ckit.h - Tiny kit of essentials for C projects github.com/RandyGaul/ckit…

@icculus Congrats!

SDL3 is officially released! 3.2.0 is our first stable build. Go get it and have a ton of fun with it! github.com/libsdl-org/SDL…