TRELLIS running on AMD GPUs with ROCm - Image to 3D Asset Generation
This is a fork of Microsoft TRELLIS modified to run on AMD consumer GPUs (tested on RX 7800 XT with ROCm 6.4.2).
| Feature | Status | Timing |
|---|---|---|
| ✅ 3D Model Generation | Working | ~45 seconds |
| ✅ Gaussian Splatting | Working (145+ it/s) | ~30 seconds |
| ✅ Gaussian Export (.ply) | Working | Instant |
| ✅ Mesh Extraction | Working | ~60 seconds |
| ✅ GLB Export with Textures | Working | 5-10 minutes |
⚠️ GLB Export Takes 5-10 Minutes: This is normal! The console will show progress through 5 steps. Your system will be under heavy load during texture baking - this is expected.
- AMD GPU (tested: RX 7800 XT, RDNA3)
- ROCm 6.4+
- Python 3.10+
- ~16GB VRAM recommended
Install libsparsehash-dev(required for building torchsparse)
Ubuntu/Debian:
sudo apt-get install libsparsehash-devFedora:
sudo dnf install sparsehash-develArch Linux
sudo pacman -S google-sparsehash# Clone the repository
git clone https://github.com/CalebisGross/TRELLIS-AMD
cd TRELLIS-AMD
# Run the installation script
chmod +x install_amd.sh
./install_amd.sh
# Activate environment and run
source .venv/bin/activate
ATTN_BACKEND=sdpa XFORMERS_DISABLED=1 SPARSE_BACKEND=torchsparse python app.pyThen open http://localhost:7860 in your browser.
| Extension | Modification |
|---|---|
| nvdiffrast-hip | AMD-safe coarse rasterizer, HIP warp intrinsic macros |
| diff-gaussian-rasterization | Manual HIP build script, buffer initialization fixes |
| torchsparse | Built with FORCE_CUDA=1 for HIP GPU backend |
- Switched to OpenGL rasterization backend (avoids HIP rasterizer bugs)
- Disabled
fill_holesin mesh postprocessing (avoids visibility check issues) - Added progress logging for GLB export
| Operation | Expected Time | Notes |
|---|---|---|
| 3D Generation (Sampling) | ~45s | 12 steps of diffusion |
| Gaussian Export | Instant | Saves .ply file |
| GLB Export | 5-10 min | Heavy CPU+GPU load is normal |
The GLB export shows progress in console:
[GLB Export] Starting GLB extraction (this takes 5-10 minutes)...
[GLB Export] Step 1/5: Mesh postprocessing...
[GLB Export] Step 2/5: UV parametrization...
[GLB Export] Step 3/5: Rendering multiview observations (100 views)...
[GLB Export] Step 4/5: Baking texture (2500 optimization steps)...
[GLB Export] Step 5/5: Finalizing GLB mesh...
[GLB Export] Complete!
- Mesh Preview: May show grey - the actual export works correctly
- fill_holes Disabled: Small holes in meshes may not be filled
- Performance: Simplified coarse rasterizer is slower than NVIDIA-optimized version
Ensure you're using ROCm 6.4+ and PyTorch built for ROCm.
Check that fill_holes=False is set in trellis/utils/postprocessing_utils.py.
Make sure you're using the AMD-modified extensions in this repo, not the original CUDA ones.
Rebuild with: cd extensions/torchsparse && CUDA_HOME=/opt/rocm FORCE_CUDA=1 pip install . --no-build-isolation
- Original TRELLIS by Microsoft
- nvdiffrast by NVIDIA
- AMD GPU modifications developed through extensive debugging of HIP compatibility issues
See original licenses for TRELLIS, nvdiffrast, and diff-gaussian-rasterization.
