TRELLIS-AMD

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 7.2.1, torch 2.10.0+rocm7.0).

Status (May 2026): Fully operational on RX 7800 XT. Tested end-to-end: image → 3D asset → textured GLB, including mesh rendering, hole filling, and texture baking on a 16 GB consumer card. The multi-month rasterizer investigation that unblocked this is documented in experiments/raster/.

Features

⚠️ 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.

Requirements

• AMD GPU (tested: RX 7800 XT, RDNA3 / gfx1101)
• ROCm 7.0+ (tested on system ROCm 7.2.1, torch 2.10.0+rocm7.0)
• Python 3.10+
• 16 GB VRAM (the pipeline is split into staged phases to fit; see example.py)

Quick Start

Install libsparsehash-dev(required for building torchsparse)

Ubuntu/Debian:

sudo apt-get install libsparsehash-dev

Fedora:

sudo dnf install sparsehash-devel

Arch 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 \
  TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL=1 python app.py

Then open http://localhost:7860 in your browser.

What's Different from Original TRELLIS?

Custom Extensions (AMD-compatible)

Application Modifications

• HIP rasterizer (CoarseRaster + FineRaster) bounds-check fix for the triHeader[i].misc OOB on RDNA3 — see experiments/raster/findings.md
• _fill_holes pole-clamps the Hammersley camera distribution so views directly above/below the mesh don't NaN view_look_at and hang coarseRaster (default num_views also dropped 1000 → 100 for speed)
• Added progress logging for GLB export
• example.py splits the pipeline into staged phases, moving idle submodels to CPU between phases so the full run fits in 16 GB VRAM

Processing Time Reference

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!

Known Limitations

1. Performance: Coarse rasterizer is serialized and slower than NVIDIA's warp-parallel version
2. ~7% silent triangle culls: The Bug 6 bounds-check fix culls triangles with an out-of-range triHeader[i].misc from triangleSetup. Visual impact is small but the underlying invariant violation is unresolved. See experiments/raster/findings.md for the Phase C root-cause hypothesis.
3. fill_holes uses 100 views, not 1000: TRELLIS upstream rasterizes 1000 Hammersley-distributed views to detect invisible faces. We clamp views away from the world-up poles (otherwise the HIP rasterizer hangs on degenerate view matrices) and use 100 views. Hole detection quality is visually indistinguishable, and step 1 of GLB extract is now ~10x faster.

Troubleshooting

GPU Hang/Crash

Ensure you're using ROCm 7.0+ and PyTorch built for ROCm (torch 2.10.0+rocm7.0 or newer is recommended).

Empty Mesh

Confirm the input image actually has a foreground subject after rembg background removal. If so, raise the Mesh Simplify slider toward 0 in the UI to keep more triangles, or pass simplify=0.0 to to_glb().

CUDA Symbol Errors

Make sure you're using the AMD-modified extensions in this repo, not the original CUDA ones.

torchsparse "no attribute" Error

Rebuild with: cd extensions/torchsparse && CUDA_HOME=/opt/rocm FORCE_CUDA=1 pip install . --no-build-isolation

Credits

• Original TRELLIS by Microsoft
• nvdiffrast by NVIDIA
• AMD GPU modifications developed through extensive debugging of HIP compatibility issues

License

See original licenses for TRELLIS, nvdiffrast, and diff-gaussian-rasterization.