DreamSampler: Unifying Diffusion Sampling and Score Distillation for Image Manipulation

arXiv:2403.11415v1 Announce Type: cross
Abstract: Reverse sampling and score-distillation have emerged as main workhorses in recent years for image manipulation using latent diffusion models (LDMs). While reverse diffusion sampling often requires adjustments of LDM architecture or feature engineering, score distillation offers a simple yet powerful model-agnostic approach, but it is often prone to mode-collapsing. To address these limitations and leverage the strengths of both approaches, here we introduce a novel framework called {\em DreamSampler}, which seamlessly integrates these two distinct …

abstract architecture arxiv cs.ai cs.cv cs.lg diffusion diffusion models distillation engineering feature feature engineering image latent diffusion models ldm manipulation model-agnostic sampling simple type