Should you replace x/y with x*(1.0/y) for shader optimization?
If "y" is a constant, it makes no difference. Modern compilers optimize this either way.
If "y" is not constant, x*(1.0/y) may be marginally faster on some hardware. Better to compute the reciprocal if used a lot.
A provocation: if this GLSL shader was one shoted with a skills.md that requests to impersonate me, and indeed does pull my work on shapes, lighting, animation and delivers my style, maths, techniques and conventions (to the variable name)... isn't this actually, MY brain's work?
Should I claim it as mine?
Claude only did the mechanical/easy part of immitating my usual moves to put this together.
Video 2 GLSL
I gave Claude Opus 5.5 a 20-second video clip. It rebuilt the whole thing as one GLSL fragment shader: 156 KB of code, no textures, no video frames. Just math, looping seamlessly
If you are new to procedural generation within game dev, here is a great starting point on how to leverage noise!
Community note
This infographic is AI-generated and mislabels several noise functions and their visual outputs. The post author acknowledged the graphic was generated by AI: x.com/slashskill_/st…
A breakdown of the inaccuracies is detailed here: x.com/XorDev/status/…
Never heard of Billow noise and hybrid is silly because many of these are already hybrids of each other.
If you want to learn real noise functions, read my tutorial please
mini.gmshaders.com/p/gm-shad…
Here's an actually good AI summary that adds some detail and clarification.
My only nitpick is that what most people call a hash, is not actually based on a hash function, and would be better described as white noise or pseudo-random