AniGen-mac Turns A Photo Into A Rigged 3D Character On Apple Silicon

AniGen-mac is a new port that brings the AniGen 3D asset generator to Apple Silicon Macs. It creates a fully rigged, animate-ready 3D model from a single input image. The entire inference pipeline now runs on Mac GPUs using Metal Performance Shaders instead of CUDA.
Developer Pawel Mazurkiewicz forked the original codebase and replaced the CUDA-only libraries with Metal equivalents. The dev also added a routine that automatically reconnects detached skeleton bones into a single tree for smooth animation. This fork supports inference only; the training stage still requires an NVIDIA GPU.
Key features of the port
- Single image to fully rigged 3D mesh.
- Runs natively on Apple Silicon Macs.
- Automatic skeleton reconnection for clean animation.
- Adjustable mesh simplification and texture resolution.
- fp32 precision ensures stable inference pipeline.
- Built-in texturing with fast projective bake mode.
- Pretrained weights downloaded automatically on first run.
Mac-based 3D artists and hobbyists can generate game-ready rigs locally without uploading images to cloud services. Small studios benefit from rapid prototyping of animated characters for simulations or games entirely on premises. Privacy-focused professionals gain a self-contained pipeline that keeps sensitive concept images offline.
Developer notes and limitations
The port is inference-only, so training remains exclusive to Linux machines with an NVIDIA GPU. Setup demands macOS Tahoe or newer, a full Xcode 26 installation, and Python 3.10 for the custom Metal extensions. According to the developer, mesh fidelity is not as high as in other single-image tools like Pixal3D, but the automatic rigging makes it fun to explore.
“The quality of the model is not terribly striking (far worse than Pixal), but I figured it's still pretty fun to explore - as it can be animated just as soon as it gets generated, which by the way takes about 3,5 minutes on my Mac (M5 Max 128GB 40c GPU).” — Source: Reddit