Fix v_mean3d in project_g and v_conic calculation in *rasterize_backward #139
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ze_backward_kernel
oseiskar
added a commit
to SpectacularAI/gsplat
that referenced
this pull request
Mar 19, 2024
…ze_backward_kernel (nerfstudio-project#139) Ported the relevant part of commit 94cbd12 to this branch. Co-authored-by: J.Y <132313008+jb-ye@users.noreply.github.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Previous #136 doesn't fully solve the problem and in fact caused the splatfacto performance degradation. This is because there are two bugs that canceling each other in training GS
Effectively they cancel out each other to some degree in splafacto when updating cov2d.
By only fixing the first bug but not the second, the performance of splatfacto significantly degraded. In an example dataset ichsan2895 provided me, PSNR degrades from 25.97 to 25.40. Post this PR, the PSNR performance is restored.
There is a third bug in updating v_mean3d which @vye16 mentioned to me earlier. This change also fix the bug and no side effect is observed (either improvement or regression).