shiimizu/ComfyUI-TiledDiffusion
A ComfyUI extension enabling high-resolution AI image generation via tiled diffusion and optimized VAE processing.

Velocity · 7d
+0.6
★ / day
Trend
→steady
star history
This extension implements tiled diffusion techniques to generate and upscale large images using diffusion models while managing VRAM constraints. It reproduces MultiDiffusion and Mixture of Diffusers methods, along with a tiled VAE algorithm for memory-efficient encoding and decoding. The tool supports SD 1.x, 2.x, SDXL, SD3, and FLUX models with ControlNet integration.