← Back to publications
2023 · Preprint

ReGeneration Learning of Diffusion Models with Rich Prompts for Zero-Shot Image Translation

Yupei Lin, Sen Zhang, Xiaojun Yang, Xiao Wang, Yukai Shi
arXiv:2305.04651
Figure 1 — ReGeneration Learning of Diffusion Models with Rich Prompts for Zero-Shot Image Translation
Figure 1. ReGeneration learning. In contrast to previous work, regenerative learning involves learning from the input X to a more concise representation Y ′, and subsequently learning from Y ′ to generate the final output Y .
Overview

Explores regeneration learning for zero-shot image translation, using cross-attention guidance and cooperative updates to preserve source-image structure without additional model training.

Paper / Publisher ↗ PDF ↗