Post #1899192
2026-04-30 15:29 UTC
New blog post!
In "Quantizing tangent frames", we look at various established methods to represent tangent frames in the vertex data, squeeze a few variants into 32 bits per vertex and look at the resulting precision.
https://zeux.io/2026/04/30/quantizing-tangent-frames/
Retoot, like and subscribe!
Replies (2)
-
@nothings@mastodon.gamedev.place 2026-04-30 17:44
@zeux@mastodon.gamedev.place Why is every production render in existence throwing away bitangents and breaking bump-mapping on UV skewing, instead of throwing away NORMAL and reconstructing THAT? It's the one that's actually mathematically supposed to be perpendicular to the other two.
-
@morten_skaaning@mastodon.gamedev.place 2026-05-01 08:54
@zeux@mastodon.gamedev.place nice article. I'm just talking out of my ass here, but would it make sense to do a "successive refinement" of the vector encoding? I.e. first use 5 bits per vector element to encode, but then decode that encoding and subtract the original vector. That would give you 3 bits per element to adjust the future decoded normal. Just a non graphics programmer thought.