RAM is the primary limit on maximum project size. GPU VRAM affects dense reconstruction speed.
| Component | Basic (< 500 images) | Advanced (1,500+ images) |
|---|---|---|
| CPU | 4–12 cores, 2.0 GHz+ | i7/i9 or Ryzen 7/9, 6–32 cores, 3.0 GHz+ |
| RAM | 16–32 GB | 32–128 GB |
| GPU | NVIDIA RTX 2060+ (CUDA) | NVIDIA RTX 3080/4070+ (CUDA) |
| Storage | SSD strongly recommended | |
Operating system
- Windows: Windows 10 or 11 (64-bit) — primary target.
- Linux: Terapetra AppImage on supported distributions (filename listed on the release notes page).
Hardware
- GPU: NVIDIA with CUDA strongly recommended for dense reconstruction.
- RAM: 8 GB minimum; 16 GB+ recommended for typical medium datasets.
- Storage: SSD strongly recommended for image sets and dense outputs.
Software (source installation)
- Python: 3.10+ via Terapetra engine conda environment
terapetra-engine(environment.yml). - External tools: COLMAP, GDAL, PDAL on
PATH(seeTERAPETRA_COLMAP_BINif needed). - Mesh: Open3D installed with Terapetra (
pip install -e ".[dev,gui,pyvista]"optional viewer extras).
Download Terapetra
See release notes on this site for installers, checksums, and version details.