magic-research
Challenges with Environment and Package Installation for magic-animate
Dear magic-animate Team,
I am experiencing difficulties while setting up the magic-animate environment. Here are the specific challenges:
- Conda Environment Setup Issues: While creating the environment with environment.yaml, I encountered ResolvePackageNotFound errors for several packages:
- openssl==1.1.1l=h7f98852_0
- jupyter_core==4.12.0=py38h578d9bd_0
- wheel==0.38.4=py38h06a4308_0
- xz==5.4.2=h5eee18b_0
- pip==23.2.1=py38h06a4308_0
- tornado==6.1=py38h0a891b7_3
- ca-certificates==2023.7.22=hbcca054_0
- libgcc-ng==11.2.0=h1234567_1
- zeromq==4.3.4=h9c3ff4c_1
- python==3.8.5=h7579374_1
- readline==8.2=h5eee18b_0
- ld_impl_linux-64==2.38=h1181459_1
- libsodium==1.0.18=h36c2ea0_1
- tk==8.6.12=h1ccaba5_0
- debugpy==1.6.7=py38h6a678d5_0
- libstdcxx-ng==11.2.0=h1234567_1
- _openmp_mutex==5.1=1_gnu
- libgomp==11.2.0=h1234567_1
- libffi==3.3=he6710b0_2
- setuptools==68.0.0=py38h06a4308_0
- pyzmq==25.1.0=py38h6a678d5_0
- sqlite==3.41.2=h5eee18b_0
- zlib==1.2.13=h5eee18b_0
- ncurses==6.4=h6a678d5_0
- pip Installation Problems: Additionally, attempting to install via pip3 install -r requirements.txt led to errors, specifically with nvidia-cudnn-cu11==8.5.0.96 not being available.
These issues might be related to package versions or Windows system compatibility. Could you provide updated setup instructions or any assistance for Windows users?
Thank you for your support.
i spent literally 24 hours to make auto installer for both dense pose maker and magic animate > https://github.com/magic-research/magic-animate/issues/44
it is really hard to make
This one worked for me with Anaconda3. Windows 10, Xeon E3-1270 v5, 16GB Ram, RTX 2060 12GB VRam.
https://github.com/EKI-INDRADI/magic-animate-for-windows-rnd-20231206
Make sure you have all these installed and the proper environment variable paths setup.
CUDA SDK 11.8 + WIN ENV (LATEST) CUDA CUDNN 8.9 + WIN ENV (LATEST) FFMPEG 6 + WIN ENV (LATEST) ANACONDA GIT
git clone [[email protected]:EKI-INDRADI/magic-animate-for-windows-rnd-20231206.git conda create -n m_animate_for_win python=3.10 -y conda activate m_animate_for_win pip install -r requirements-windows.txt
Make sure you have the proper folder structure and files particularly the Stable-diffusion-v1-5 folder. Its about 13GBs with a 2GB SD1.5 model.
magic-animate |----pretrained_models (create folder pretrained_models) |----MagicAnimate (git lfs clone https://huggingface.co/zcxu-eric/MagicAnimate) |----appearance_encoder |----diffusion_pytorch_model.safetensors |----config.json |----densepose_controlnet |----diffusion_pytorch_model.safetensors |----config.json |----temporal_attention |----temporal_attention.ckpt |----sd-vae-ft-mse (manual download https://huggingface.co/stabilityai/sd-vae-ft-mse) |----config.json |----diffusion_pytorch_model.safetensors |----stable-diffusion-v1-5 (manual download https://huggingface.co/runwayml/stable-diffusion-v1-5, dont donload all 100GB+) |----scheduler |----scheduler_config.json |----text_encoder |----config.json |----pytorch_model.bin |----tokenizer (all) |----unet |----diffusion_pytorch_model.bin |----config.json |----v1-5-pruned-emaonly.safetensors |----...
how to run
python -m demo.gradio_animate
Good luck
edit: check the link for the right folder structure. When I posed it made everything align right.
