Local InstantMesh Tiger

Following the instructions in this repo https:// github.com/TencentARC/InstantMesh very closely, I created and activated a conda environment using minconda prompt in windows. Then separately cloned the repo in C:// using git bash. I changed directory to the root of the cloned repo in miniconda and did the following: conda create --name instantmesh python=3.10 conda activate instantmesh pip install -U pip # Ensure Ninja is installed conda install Ninja # Install the correct version of CUDA conda install cuda -c nvidia/label/cuda-12.1.0 # Install PyTorch and xformers # You may need to install another xformers version if you use a different PyTorch version pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https:// download.pytorch.org/whl/cu121 pip install xformers==0.0.22.post7 #then download and put this triton wheel in the root of the InstantMesh repo https:// github.com/PrashantSaikia/Triton-for-Windows/blob/main/triton-2.0.0-cp310-cp310-win_amd64.whl # Install other requirements pip install -r requirements.txt In the same minconda command prompt, update huggingface hub. This makes a mismatch between huggingface hub and tokenizers, but tokenizers is not required since there is no text to 3d model generation. Now run app.py in the prompt then it will download several models and diffusion models. I changed the default port to 7860 in the code for app.py. Once the server runs, copy this to your browser url bar: http:// 127.0.0.1:7860/ When you want to run it locally, just activate the conda environment through miniconda command prompt, cd to the directory containing the repo i.e. 'path/to/your/InstantMesh', and run python app.py . I hope this video helps as InstantMesh is my favorite 3d reconstruction model. It has applications in the game dev world, architecture, movies, and other 3d modeling use cases. Cheers!