TRELLIS.2 Image-to-3D Model Ported to Apple Silicon Macs
Original: Show HN: Run TRELLIS.2 Image-to-3D generation natively on Apple Silicon
Why This Matters
Makes advanced 3D generation accessible to Mac users without expensive GPU hardware
Developer shivampkumar has ported Microsoft's TRELLIS.2 image-to-3D generation model to run natively on Apple Silicon Macs using PyTorch MPS, eliminating the need for NVIDIA GPUs and generating 400K+ vertex meshes in ~3.5 minutes on M4 Pro.
The port converts Microsoft's state-of-the-art TRELLIS.2 model from CUDA-only to Apple Silicon compatibility via PyTorch MPS backend. The system generates detailed 3D meshes with over 400,000 vertices from single images in approximately 3.5 minutes on M4 Pro hardware. Output includes textured OBJ and GLB files with PBR materials ready for 3D applications. Requirements include macOS on Apple Silicon (M1 or later), Python 3.11+, 24GB+ unified memory, and 15GB disk space for model weights. The GitHub repository provides setup scripts and generation tools for local deployment without cloud dependencies.