Closed weihan1 closed 6 days ago
ok nvm solved, i basically had to remove the commit hash in the toml file prior to installation
Hello, I also encountered a similar issue. Could you please tell me how you resolved it? Thank you.
what worked for me was removing the commit hashes for diff-gaussian-rasterization and simple-knn in the toml file:
[build-system] requires = ["setuptools>=61.0"] build-backend = "setuptools.build_meta"
[tool.setuptools.packages.find] include = ["internal", "utils"]
[project] name = "gaussian-splatting-lightning" dynamic = ["version"] requires-python = ">=3.8" dependencies = [ "lightning[pytorch-extra]==2.0.9.post0", "pytorch-lightning==2.3.", "splines==0.3.0", "plyfile==0.8.1", "tensorboard", "wandb", "tqdm", "viser==0.2.3", "opencv-python==4.10.", "matplotlib", "mediapy==1.2.2", "diff-gaussian-rasterization @ git+https://github.com/graphdeco-inria/diff-gaussian-rasterization.git", "simple-knn @ git+https://gitlab.inria.fr/bkerbl/simple-knn.git"]
[project.scripts] gs-fit = "internal.entrypoints.gspl:cli_fit" gs-val = "internal.entrypoints.gspl:cli_val" gs-test = "internal.entrypoints.gspl:cli_test" gs-predict = "internal.entrypoints.gspl:cli_predict"
segany-fit = "internal.entrypoints.seganygs:cli_fit" segany-val = "internal.entrypoints.seganygs:cli_val" segany-test = "internal.entrypoints.seganygs:cli_test" segany-predict = "internal.entrypoints.seganygs:cli_predict"
gs-viewer = "internal.entrypoints.viewer:cli"
Thank you for your response. I’m sorry, but this doesn’t help me. My original server was CUDA 11.5, and everything went smoothly when I switched to a server with CUDA 12.4. I still appreciate your reply.
Hello,
Thanks for the great work! I have encountered this issue:
ImportError: /scratch/ondemand28/weihanluo/miniconda3/envs/gspl/lib/python3.9/site-packages/diff_gaussian_rasterization/_C.cpython-39-x86_64-linux-gnu.so: undefined symbol: _ZN3c106detail19maybe_wrap_dim_slowEllb
when running python viewer.py <4dgs_output_dir> --vanilla_gs4d
I followed the instructions from the readme, my pytorch version is 2.0.1 with cuda 11.7 and pip install -e . ran successfully. Any help would be greatly appreciated 🙏