Anttwo / SuGaR

[CVPR 2024] Official PyTorch implementation of SuGaR: Surface-Aligned Gaussian Splatting for Efficient 3D Mesh Reconstruction and High-Quality Mesh Rendering

Home Page:https://anttwo.github.io/sugar/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete.

Runyu-Zhou05 opened this issue · comments

System: Ubuntu 22.04.4 LTS
GPU: NVIDIA RTX 3090

Dataset: tandt_db
Scene: truck
GS Iterations: 30000
Command: python train.py -s ../../gs-repo/tandt/truck -c ./gsoutputs/tandt_truck_long/ -r sdf --refinement_time long --iteration_to_load 30000

A part of of the output:

--------------------
Loading the initial 3DGS model from path ./gsoutputs/tandt_truck_long/...
Found image extension .jpg
219 training images detected.
The model has been trained for 30000 steps.

Loading the coarse SuGaR model from path ./output/coarse/truck/sugarcoarse_3Dgs30000_sdfestim02_sdfnorm02/15000.pt...
Use min to initialize scales.
Initialized radiuses for 3D Gauss Rasterizer
Coarse model loaded.
Coarse model parameters:
_points
torch.Size([709310, 3])
True
all_densities
torch.Size([709310, 1])
True
_scales
torch.Size([709310, 3])
True
_quaternions
torch.Size([709310, 4])
True
_sh_coordinates_dc
torch.Size([709310, 1, 3])
True
_sh_coordinates_rest
torch.Size([709310, 15, 3])
True
Number of gaussians: 709310
Opacities min/max/mean: tensor(1.0736e-13, device='cuda:0') tensor(1., device='cuda:0') tensor(0.9659, device='cuda:0')
Quantile 0.0: 1.0736448893561984e-13
Quantile 0.1: 0.975199282169342
Quantile 0.2: 0.9996374845504761
Quantile 0.3: 0.999923825263977
Quantile 0.4: 0.9999750852584839
Quantile 0.5: 0.9999910593032837
Quantile 0.6: 0.9999967813491821
Quantile 0.7: 0.9999988079071045
Quantile 0.8: 0.9999996423721313
Quantile 0.9: 1.0

Starting pruning low opacity gaussians...
WARNING! During optimization, you should use a densifier to prune low opacity points.
This function does not preserve the state of an optimizer, and sets requires_grad=False to all parameters.
Number of gaussians left: 691576
Opacities min/max/mean: tensor(0.5000, device='cuda:0') tensor(1., device='cuda:0') tensor(0.9842, device='cuda:0')
Quantile 0.0: 0.5000001788139343
Quantile 0.1: 0.9944981336593628
Quantile 0.2: 0.9997538924217224
Quantile 0.3: 0.9999387264251709
Quantile 0.4: 0.9999786615371704
Quantile 0.5: 0.9999921321868896
Quantile 0.6: 0.9999970197677612
Quantile 0.7: 0.999998927116394
Quantile 0.8: 0.9999996423721313
Quantile 0.9: 1.0
Processing frame 0/219...
Current point cloud for level 0.3 has 0 points.
Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase. This caused an overflow, meaning output may be incomplete. To solve, try increasing max_faces_per_bin / max_points_per_bin, decreasing bin_size, or setting bin_size to 0 to use the naive rasterization.Bin size was too small in the coarse rasterization phase.