camenduru

💃 Image to VRM + VRMA (GLB2VRM, VRMExMR) is now fully integrated into OpenCartridge. 🥳 🧬camenduru/OpenCartridge Also added a new model cartridge: UniMate Motion. 🥳 🦋 One Unified Model to Animate Diverse Skeletons Thanks to ❤ 🧬Friedrich-M/UniMate t.co

💃 VRMExMR 🕺 🧬camenduru/VRMExMR 🐳docker run -d --name vrmexmr --restart unless-stopped --gpus all --shm-size=8g -p 8010:8010 -v vrmexmr-data:/app/outputs camenduru/vrmexmr Thanks to ❤ Tom Wehrbein, Bodo Rosenhahn 🩰 ExMR: Extremity Motion Refinement for Monocular Human Motion Capture 🧬twehrbein/ExMR 🌐twehrbein.github.io 📄arxiv.org

👻🦇 VRMHairMaker 🎃💀 🧬camenduru/VRMHairMaker 🐳docker run -d --name vrm-hairmaker ---gpus all -p 8300:8300 camenduru/vrmhairmaker Thanks to ❤ Yuze He, Yixiang Wei, Wang Zhao, Chuhao Chen, Yanning Zhou, Yong-Jin Liu 💈 HairDeck: Animation-Ready Stylized Hair Cards from a Single Image 🧬hyz317/HairDeck 🌐hyzcluster.github.io 📄arxiv.org

Free Local Studio Turns a Few Unposed Photos Into a Textured 3D Model UniForge3D Studio, a local web app built around UniForge3D, the new open-source model from Peking University, HKUST and Math Magic. You drop in two or more photos of an object from any angles, no camera positions needed, and get back a textured GLB. The model keeps what the photos show and generates the hidden sides in the same 3D space, so the back stays true to the real object. What's in the box: - One command: docker run --gpus all -p 8091:8091 camenduru/uniforge3d-studio, then open localhost:8091 in your browser - Fully offline: every model is already inside the image, so there's no Hugging Face login and nothing gets downloaded or sent anywhere while it runs - Simple web UI: upload photos (PNG, JPG or WebP), pick a seed, hit Generate and check the result in the built-in 3D viewer. Background removal runs automatically - Optimize for games: meshoptimizer's gltfpack is built in, so you can cut the triangle count (to 25% by default), compress the mesh and convert textures to WebP - Low-VRAM mode: moves some stages to the CPU so a run fits on a 24 GB card The catch is size. The image is a 36 GB download (around 55 GB unpacked), it needs an NVIDIA GPU with about 18 GB of free VRAM, and one model takes roughly 10 minutes on an RTX 3090, coming out near 50 MB before optimization. It's a community wrapper by camenduru rather than the authors' release, and some bundled weights, like the RMBG-2.0 background remover, are non-commercial, so commercial use is still unclear. No big GPU? There's a community demo on Hugging Face. github: camenduru/UniForge3D

Malfoid lives on your desktop now. Good luck. 🧬code: camenduru/VRMPet 🏎️engine: turanszkij/WickedEngine 🀥vrm: ko-fi.com t.co

🎛 Design your favorite character's voice, personality, and appearance then give them a call. Or generate any character's voice with personality and acting skills. 🌏 Works in any language. Thanks to ❤ @FishAudio for sponsoring this project. 🧬code: camenduru/fish-audio-drama-3 🐠try(password=fish): fish-audio-drama-3.vercel.app 🐟page: fish.audio 📦model: huggingface.co