Cascadeur: NVIDIA Kimodo script/plugin [v2 UPDATE + Patch 1.0]

Patch 1.0: - Multi-character processing - Fixed issues with shoulders going higher the more you process with Kimodo - Fixed original Kimodo template - Fixed issue when first constraint pose is far from 0,0,0 DOWNLOAD LINK (v2, has all models, plug-and-play) https://drive.google.com/file/d/1sgxquPqi0HcA-TijPQKeKjekl_PwNHHm/view?usp=sharing Patch 1.0 link: https://drive.google.com/file/d/1dAeictUMVztvncjjWhuMLQjcs0XVsLCW/view - Includes EVERYTHING and has a 10GB VRAM mode for LLAMA (8-bit quantized + CPU offload) - Slightly worse than the full 22GB VRAM eating LLAMA version, but can be run on many 12GB+ VRAM NVIDIA GPUs. - Extract to any folder, and run Install.bat to install WSL, adjust path info and it will copy the script to Cascadeur folder - Check Install.txt for more information - It takes some time to Start Kimodo (~1-3 min), so don't worry - VRAM USAGE (LLAMA ON): ~17-20 GB (for pose constraints and prompts) - VRAM USAGE (LLAMA 8-bit + CPU offload ): Max 10 GB (for pose constraints and prompts) - VRAM USAGE (LLAMA OFF): from my usage... ~2-4 GB (for pose constraints only) . . . OLD VERSION DOWNLOAD LINK (in case for some reason you have issues?...): https://drive.google.com/file/d/1g-78PJeo8gIGHs4Y3i8cJgsDqdEZ-b80/view?usp=sharing - Includes everything besides Meta LLAMA 3 8B-Instruct (download here, requires personal info: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) (download all files and put them into "\KimodoCascadeur\text-encoders\meta-llama\Meta-Llama-3-8B-Instruct\", make sure file structure is correct) - Its vibecoded but tested slop, please DM me for any errors or vibefix it yourself ( ಠل͟ಠ)