task-lora

Tag

Cards List
#task-lora

From RGB Generation to Dense Field Readout: Pixel-Space Dense Prediction with Text-to-Image Models

Hugging Face Daily Papers ↗ · 2026-07-09 Cached

ReChannel adapts pretrained diffusion transformers (e.g., FLUX-Klein) for dense prediction tasks by mapping tokens directly to pixel-space patches, achieving state-of-the-art results on trimap-free matting, KITTI depth, and referring segmentation with minimal additional parameters.

0 favorites 0 likes
← Back to home

Submit Feedback