docs: Clouds example #1655
docs: Clouds example #1655
Conversation
|
pkg.pr.new packages benchmark commit |
|
Reverted perlin noise, due to performance issues (both impl with texture and without), even when using cache. |
|
Once approved, I will add example resolution test for it. |
| export const raymarchSlot = tgpu.slot<TgpuFn>(); | ||
| export const sampledViewSlot = tgpu.slot<TgpuSampledTexture<'2d', d.F32>>(); | ||
| export const samplerSlot = tgpu.slot<TgpuSampler>(); | ||
| export const timeAccess = tgpu['~unstable'].accessor(d.f32); | ||
| export const resolutionAccess = tgpu['~unstable'].accessor(d.vec2f); |
There was a problem hiding this comment.
Why not use these resources directly? What is the advantage of using these? Is this for abstraction?
There was a problem hiding this comment.
Pretty much. I briefly tried to introduce cloud profiles into the example, but all my endeavours came out looking ugly. I left the abstraction for future extensibility and better file fragmentation
| const lit = d.vec3f( | ||
| albedo.x * lighting.x, | ||
| albedo.y * lighting.y, | ||
| albedo.z * lighting.z, | ||
| ); | ||
| const premul = d.vec4f( | ||
| lit.x * density, | ||
| lit.y * density, | ||
| lit.z * density, | ||
| density, | ||
| ); | ||
|
|
||
| res = std.add(res, std.mul(premul, LIGHT_ABSORPTION - res.w)); |
There was a problem hiding this comment.
nit
| const lit = d.vec3f( | |
| albedo.x * lighting.x, | |
| albedo.y * lighting.y, | |
| albedo.z * lighting.z, | |
| ); | |
| const premul = d.vec4f( | |
| lit.x * density, | |
| lit.y * density, | |
| lit.z * density, | |
| density, | |
| ); | |
| res = std.add(res, std.mul(premul, LIGHT_ABSORPTION - res.w)); | |
| const lit = albedo.mul(lighting); | |
| const premul = d.vec4f(lit, 1).mul(density); | |
| res = res.add(premul.mul(LIGHT_ABSORPTION - res.w)); |
There was a problem hiding this comment.
Sometimes there are clearly visible vertical or horizontal divisions, do you know where they come from and if they are fixable?
Screen.Recording.2025-09-02.at.16.33.23.mov