DesignEdit: Multi-Layered Latent Decomposition and Fusion for Unified & Accurate Image Editing

arXiv:2403.14487v1 Announce Type: new
Abstract: Recently, how to achieve precise image editing has attracted increasing attention, especially given the remarkable success of text-to-image generation models. To unify various spatial-aware image editing abilities into one framework, we adopt the concept of layers from the design domain to manipulate objects flexibly with various operations. The key insight is to transform the spatial-aware image editing task into a combination of two sub-tasks: multi-layered latent decomposition and multi-layered latent fusion. First, we segment the …

abstract arxiv attention concept cs.cv design domain editing framework fusion image image generation image generation models objects spatial success text text-to-image type