Guys,
Thanks for your excellent work, really impressive!
Do you have any recommendations / ideas on how can I decrease # of splats generated for low-end devices? I see flame_subdivide parameter in inference and it seems to be working (at least it says that 5k splats are generated), however I cannot export the avatar generated using provided Blender templates. As I understand this is because of decreased. number of edges / vertices.
Do you have by chance template for original flame topology with 5k splats generated?