search

graphdeco-inria / hierarchical-3d-gaussians

Official implementation of the SIGGRAPH 2024 paper "A Hierarchical 3D Gaussian Representation for Real-Time Rendering of Very Large Datasets"
Other
932 stars 87 forks source link

error in my dataset when use hierarchy-rasterizer #28

Open vanlic opened 3 months ago

vanlic commented 3 months ago

Error: File "/home/wlc/anaconda3/envs/hierarchical_3d_gaussians/lib/python3.12/site-packages/diff_gaussian_rasterization/__init__.py", line 84, in forward

num_rendered, color, radii, geomBuffer, binningBuffer, imgBuffer, invdepths = _C.rasterize_gaussians(*args)

RuntimeError: numel: integer multiplication overflow

Training progress: 0%| | 0/30000 [00:00<?, ?it/s]

Error executing train_coarse: Command 'python train_coarse.py -s /home/wlc/Documents/3DGS/data/hds/camera_calibration/aligned --save_iterations -1 -i ../rectified/images --skybox_num 100000 --model_path /home/wlc/Documents/3DGS/data/hds

my environment：

mx-liu6 commented 2 months ago

I have encountered the same problem, I suspect that the depth of the sky is too large when the depth map is estimated? I'm trying to solve it now

mx-liu6 commented 2 months ago

Have you solved it? Do you have any ideas how to find the reasons? @vanlic

hanfeiwu commented 2 months ago

same problem

vanlic commented 2 months ago

I have encountered the same problem, I suspect that the depth of the sky is too large when the depth map is estimated? I'm trying to solve it now

I'm using an indoor dataset, so there shouldn't be a problem with too large depth, and I haven't found a way yet