search

fudan-zvg / 4d-gaussian-splatting

[ICLR 2024] Real-time Photorealistic Dynamic Scene Representation and Rendering with 4D Gaussian Splatting
MIT License
611 stars 44 forks source link

Issue with libopenh264.so.5 while running n3v2blender.py script #43

Open SHIBOYA opened 4 months ago

SHIBOYA commented 4 months ago

Dear Zeyu,

I am currently working with the 4D Gaussian Splatting project and encountered an issue while attempting to run the n3v2blender.py script. When I execute the script with the following command:

python scripts/n3v2blender.py /root/autodl-tmp/N3V/cut_roasted_beef

I receive the following error message:

==== running: ffmpeg -i /root/autodl-tmp/N3V/cut_roasted_beef/cam20.mp4 -start_number 0 /root/autodl-tmp/N3V/cut_roasted_beef/images//cam20_%04d.png
ffmpeg: error while loading shared libraries: libopenh264.so.5: cannot open shared object file: No such file or directory
FATAL: command failed

I have tried several solutions, including installing the openh264 library and adding the libde265 repository, but the issue persists. Here are the steps I followed:

  1. Tried to install openh264 directly using sudo apt-get install openh264, but it was not found.
  2. Added the libde265 repository and installed it manually:
    • Added the repository using sudo add-apt-repository ppa:strukturag/libde265
    • Installed openh264 using sudo apt-get install openh264
  3. Downloaded and manually installed the OpenH264 library from GitHub:
    • Downloaded the library using wget
    • Extracted and moved the library to /usr/lib
    • Updated the linker cache using sudo ldconfig

Despite these efforts, I still encounter the same error. Could you please provide guidance on how to resolve this issue or any additional steps I might be missing?

Thank you very much for your time and assistance.

Best regards!

Yujie-G commented 3 months ago

This works for me: link ; try also which ffmpeg to find the path and then deactivate if it's in /usr/lib/ dir

SHIBOYA commented 3 months ago

After install ffmpeg, I encountered an error while running train.py. The error message indicates "Could not recognize scene type!".

@ubun:~/4d-gaussian-splatting$ CUDA_VISIBLE_DEVICES=0 python train.py --config /data/shiboya/4d-gaussian-splatting/configs/dnerf/jumpingjacks.yaml
Using /data/shiboya/.cache/torch_extensions/py37_cu116 as PyTorch extensions root...
Detected CUDA files, patching ldflags
Emitting ninja build file /data/shiboya/.cache/torch_extensions/py37_cu116/diff_gaussian_rasterization/build.ninja...
Building extension module diff_gaussian_rasterization...
Allowing ninja to set a default number of workers... (overridable by setting the environment variable MAX_JOBS=N)
ninja: no work to do.
Loading extension module diff_gaussian_rasterization...
Optimizing output/dnerf/jumpingjacks
Output folder: output/dnerf/jumpingjacks [30/07 17:39:07]
Tensorboard not available: not logging progress [30/07 17:39:07]
Traceback (most recent call last):
  File "train.py", line 403, in <module>
    args.gaussian_dim, args.time_duration, args.num_pts, args.num_pts_ratio, args.rot_4d, args.force_sh_3d, args.batch_size)
  File "train.py", line 46, in training
    scene = Scene(dataset, gaussians, num_pts=num_pts, num_pts_ratio=num_pts_ratio, time_duration=time_duration)
  File "/data/shiboya/4d-gaussian-splatting/scene/__init__.py", line 52, in __init__
    assert False, "Could not recognize scene type!"
AssertionError: Could not recognize scene type!