Docs · Questions
StreamDiffusionTDInstallation
What Python and CUDA versions were required for legacy StreamDiffusionTD versions (pre-v0.2.6)?
Legacy Python & CUDA Requirements for StreamDiffusionTD
⚠️ LEGACY INFORMATION - Applies to v0.1.0 through v0.2.5 (January 2024 - February 2025)
Standard Requirements (Pre-v0.2.6)
Python Version
- Primary Recommended: Python 3.10.9
- Also Supported: Python 3.11.9 (was always supported, just not primary recommendation)
- Update in v0.2.6 (Feb 12, 2025): Python 3.11.9 became the primary recommended version for better TouchDesigner integration
CUDA Version
- Required: CUDA 11.8
- Not Compatible: CUDA 12.1+ had installation issues in early versions (v0.1.3 era)
- Issue Fixed: CUDA 12.1 installation problems resolved in later v0.1.x updates
When to Use This Information
This information is relevant if you are:
- Using StreamDiffusionTD versions v0.1.0 through v0.2.5
- Troubleshooting installation issues on older setups
- Running legacy projects that can’t be upgraded
Modern Recommendations (v0.2.6+)
- Python 3.11.9: Now primary recommendation (works inside TouchDesigner for other tools)
- CUDA 12.x: Supported in v0.2.6+, with CUDA 12.8 recommended for v0.3.0
Installation Context
The Python environment is created as a virtual environment (venv) outside of TouchDesigner. The operator handles this automatically during installation - you don’t need to configure Python inside TD itself.
Version Timeline:
- v0.1.0-v0.1.2 (Jan-Feb 2024): Python 3.10.9 + CUDA 11.8 only
- v0.1.3 (Feb 2024): Added CUDA 12.1 installation fixes
- v0.2.6 (Feb 2025): Python 3.11.9 became primary recommendation
- v0.3.0 (Nov 2025): CUDA 12.8 support added