Weird results

[StarArawn]

First off thanks for this really neat tool to pre-compute the scattering tables.

I'm seeing some issues that don't make much sense to me:

Some things I've done differently:

• I use an SRGB frame buffer. I'm not sure what three.js uses?
• I used HDR textures instead of exr as my custom engine supports loading HDR image files.
• I had to inverse the V texture coord using v = 1.0 - v.
• I use a single full-screen triangle to render everything and calculate the view direction using this code:

  mat3 invModelView = transpose(mat3(view));
  vec3 unProjected = (inverse(projection) * pos).xyz;
  v_Uv = invModelView * unProjected;

  I've used this technique successfully for cube-map rendering in other shaders.

One thing I noticed is that the resulting exr images appear to have white pixel data to the right side of the texture:

When I converted these textures to hdr they seemingly turn to black for some reason.

[danielshervheim]

Apologies for the long-delayed response. I would guess that something is going wrong in the EXR to HDR conversion process, since the textures shouldn't turn black. That said, they do contain linear data, so they should not be treated as sRGB textures. The white strip is a problem that I have encountered myself, I have not yet determined the cause of it. The first image you posted almost looks like its tone mapped incorrectly (but this might be due to treating the textures as sRGB rather than linear data)...