News for hackers. Stuff that matters.
Re:Compressing Gaussian Splats by slimbuck on Wednesday December 06, 2023 @08:21AM (#38543500)
It's definitely possible to compress/palletise the SH bands (see https://aras-p.info/blog/2023/09/27/Making-Gaussian-Splats-m... for one approach).

We just found that splats encode some view-dependent shininess anyway (for example the guitar body https://playcanvas.com/model-viewer?load=https://code.playca...) and so considered the extra data and rendering cost of SH not worth it for now.

It would be quite straightforward to add SH to the PLY file and we'll definitely consider it in future!

Re:Compressing Gaussian Splats by ladberg on Wednesday December 06, 2023 @09:16AM (#38544146)
Oh nice! I didn't realize those kinds of effects were possible without SH
Re:Compressing Gaussian Splats by jayd16 on Wednesday December 06, 2023 @11:46AM (#38546277)
Are the baked in reflections technically bug or is it fundamental in some way?
Re:Compressing Gaussian Splats by taneq on Wednesday December 06, 2023 @05:28PM (#38550374)
They’re fundamental to the scene if you recite light transport and not just geometry. Gaussian splats are just a clever way to sample a lightfield and reflections are part of the lightfield.
Re:Compressing Gaussian Splats by jayd16 on Thursday December 07, 2023 @06:10PM (#38563063)
Well it's more that I'm asking if lighting data should live entirely in the spherical harmonics. It's certainly feasible and common in the game world.

...But maybe that doesn't make sense in the GS context.

Re:Compressing Gaussian Splats by crubier on Wednesday December 06, 2023 @12:41PM (#38547097)
Yeah splats can encode view dependency without SH, but not with a correct geometry. E.g. often, you see reflections encoded in GS as just a 3d scene on the other side of the mirror. Other view dependency effect can make the surfaces "fluffy", etc. This is not great when you also try to get decent geometry.

That said, I don't think SH are the best way to represent view dependency. Other approaches using MLP or deep features seem more promising.