ComfyUI-FlowMatching-Upscaler

DyPEQwenImage

Applies DyPE-style spatial extrapolation to a Qwen Image diffusion model.

Model Patches/Unet · 1 output

FlowMatchingProgressiveUpscaler

Progressive upscaler tailored for flow-matching models. Each stage performs: 1. Latent upscaling. 2. Flow-style re-noising (linear mix with fresh noise). 3. Denoising via the configured sampler. 4. Skip residual blending to preserve coarse composition. 5. Optional dilated sampling refinement.

Latent/Upscaling · 5 outputs

FlowMatchingStage

Compact description unavailable for this node.

Latent/Upscaling · 6 outputs

FlowMatchingStageMerge

Compact description unavailable for this node.

Latent/Upscaling · 2 inputs · 1 parameter · 1 output

FlowMatchingStagePrep

Compact description unavailable for this node.

Latent/Upscaling · 4 outputs

LatentUpscaleAdvanced

Covariance-aware latent upscaler inspired by `upscaling.md`. Implements: - Whiten → upscale → recolor (global PCA/eigen whitening). - Optional moment-preserving correction to re-match mean/covariance after scaling.

Latent/Upscaling · 1 output