You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Just adding some notes regarding the spec and what should be clarified.
Judging from the description of the index buffer, it sounds like you could run into a situation where the extra skirt vertices generated client-side could push the count over 65536 which means the u16 index buffer needs to be copied over to u32 index buffer.
Don't think it's an issue in practice but something that should be clarified/addressed if the 3dtiles spec ever absorbs quantized mesh that assumes some client side post processing like skirt generation. Also pretty sure there's a reliable way to generate all 4 skirt sides of a tile using only 4 more vertices.
It doesn't mention if the edge vertices are sorted but I think it should explicitly state that they are.
The text was updated successfully, but these errors were encountered:
Just adding some notes regarding the spec and what should be clarified.
Judging from the description of the index buffer, it sounds like you could run into a situation where the extra skirt vertices generated client-side could push the count over 65536 which means the u16 index buffer needs to be copied over to u32 index buffer.
Don't think it's an issue in practice but something that should be clarified/addressed if the 3dtiles spec ever absorbs quantized mesh that assumes some client side post processing like skirt generation.
Also pretty sure there's a reliable way to generate all 4 skirt sides of a tile using only 4 more vertices.It doesn't mention if the edge vertices are sorted but I think it should explicitly state that they are.
The text was updated successfully, but these errors were encountered: