VL.StreamDiffusion
Tebjan's vvvv library for realtime image generation, with the Python stack RAM runs on.
What it is
VL.StreamDiffusion is a vvvv library by Tebjan Halm for generating images in real time with AI, together with the Python software under it (github.com/tebjan/VL.StreamDiffusion).
For us: Realtime AI Mixer (RAM) is built on it. The Python stack is Miniconda, CUDA 12.8, PyTorch and TensorRT. TensorRT engines are built per GPU on the first run, which takes a long time and a lot of disk space.
How to start
- Run the install scripts from C:\AI in order (VL.PythonNET, VL.StreamDiffusion, installer.bat).
- Run the model download script.
- First start builds the TensorRT engines.
Find it
- Repo: tebjan/VL.StreamDiffusion (branch sdxl)
- Local:
C:\AI\VL.StreamDiffusion
Ask
Armin SeltzNotion help; the Hetzner server; SpaceMusic programming- Where it lives
- RAM PCs with an NVIDIA GPU, in C:\AI\VL.StreamDiffusion.