VL.StreamDiffusion - Animated 3d scene base - 2025 05 25 22 42 27
More playing around with tebjan's VL.StreamDiffusion which enables realtime image generation natively within vvvv gamma
The image generator closely follows the source prompt image.
What can be achieved by generating this image using a 3D scene, which is animated?
A strong composition in the source image can lend to almost an infinite amount of highly usable generated imagery.
Throughout the video the prompt is generally static, with fast seed switches without interpolation.
Additional detail is enforced by using Worley noise as both texture and eventually also a flat overlay in the image prompt.
Prompt sets are taken from Branches project