mirror of
https://github.com/nerfstudio-project/nerfstudio.git
synced 2026-11-13 03:14:00 +08:00
note on hashgrid sizes
This commit is contained in:
1 parent
7ee1e31281
commit
af170840c5
1 file changed
+2
@@ -38,6 +38,7 @@ lerf_method = MethodSpecification(
|
||||
),
|
||||
model=LERFModelConfig(
|
||||
eval_num_rays_per_chunk=1 << 15,
|
||||
# NOTE: exceeding 16 layers per hashgrid causes a segfault within Tiny CUDA NN, so instead we compose multiple hashgrids together
|
||||
hashgrid_sizes=(19, 19),
|
||||
hashgrid_layers=(12, 12),
|
||||
hashgrid_resolutions=((16, 128), (128, 512)),
|
||||
@@ -88,6 +89,7 @@ lerf_method_big = MethodSpecification(
|
||||
),
|
||||
model=LERFModelConfig(
|
||||
eval_num_rays_per_chunk=1 << 15,
|
||||
# NOTE: exceeding 16 layers per hashgrid causes a segfault within Tiny CUDA NN, so instead we compose multiple hashgrids together
|
||||
hashgrid_sizes=(19, 19),
|
||||
hashgrid_layers=(16, 16),
|
||||
hashgrid_resolutions=((16, 128), (128, 512)),
|
||||
|
||||
Reference in new issue
Block a user