About the Demo Data

[fdyuandong]

Great work! Due to the complexity of the overall runtime environment, could you please provide a demo data along with the intermediate results? Thank you very much!

[egorzakharov]

Yes, we're trying to work out which demo data we can publish and how

[fdyuandong]

I’m looking forward to you releasing the demo data!

[snow-nene]

Yes, we're trying to work out which demo data we can publish and how

I20241027 10:03:24.126905 139823641604096 incremental_mapper.cc:378] => Image sees 1876 / 2744 points I20241027 10:03:24.996936 139823641604096 incremental_mapper.cc:41] Retriangulation and Global bundle adjustment W20241027 10:03:58.655370 139823641604096 levenberg_marquardt_strategy.cc:123] Linear solver failure. Failed to compute a step: CHOLMOD warning: Matrix not positive definite. W20241027 10:04:01.479920 139823641604096 levenberg_marquardt_strategy.cc:123] Linear solver failure. Failed to compute a step: CHOLMOD warning: Matrix not positive definite. W20241027 10:04:07.657245 139823641604096 levenberg_marquardt_strategy.cc:123] Linear solver failure. Failed to compute a step: CHOLMOD warning: Matrix not positive definite. W20241027 10:04:13.775213 139823641604096 levenberg_marquardt_strategy.cc:123] Linear solver failure. Failed to compute a step: CHOLMOD warning: Matrix not positive definite.

I crop your video on youtube as the demo data , and it has errors. I was wondering if there is any requirement to the video? such as resolution?

[kevinu0628]

I am implementing your codebase using frames cropped from teaser.mp4, but encountered technical obstacles during the first step in run.sh.
(base) lfl@ubun:~/lfl_dev/hair_research/GaussianHaircut$ bash run_reconstruction_cmd.sh
Optimizing /home/lfl/lfl_dev/hair_research/GaussianHaircut/data/3d_gaussian_splatting/stage1
Output folder: /home/lfl/lfl_dev/hair_research/GaussianHaircut/data/3d_gaussian_splatting/stage1 [14/03 13:40:58]
Reading camera 401/401 [14/03 13:40:59]
Loading Training Cameras [14/03 13:40:59]
100%|████████████████████████████████████████████████████████████████████████████| 128/128 [00:02<00:00, 55.61it/s]
Loading Test Cameras [14/03 13:41:01]
0it [00:00, ?it/s]
Number of points at initialisation : 78063 [14/03 13:41:01]
Training progress: 0%| | 0/30000 [00:00<?, ?it/s]Traceback (most recent call last):
File "/home/lfl/lfl_dev/hair_research/GaussianHaircut/src/train_gaussians.py", line 321, in
training(lp.extract(args), op.extract(args), pp.extract(args), args.test_iterations, args.save_iterations, args.checkpoint_iterations, args.start_checkpoint, args.debug_from)
File "/home/lfl/lfl_dev/hair_research/GaussianHaircut/src/train_gaussians.py", line 110, in training
render_pkg = render(viewpoint_cam, gaussians, pipe, background)
File "/home/lfl/lfl_dev/hair_research/GaussianHaircut/src/gaussian_renderer/init.py", line 106, in render
radii[points_mask] = _radii
RuntimeError: CUDA error: an illegal memory access was encountered
Compile with TORCH_USE_CUDA_DSA to enable device-side assertions.To ensure reproducibility, would you consider releasing the canonical demo dataset?

Your guidance would be greatly appreciated.Looking forward to your reply.