AI RESEARCH
DecQ: Detail-Condensing Queries for Enhanced Reconstruction and Generation in Representation Autoencoders
arXiv CS.CV
•
ArXi:2605.22777v1 Announce Type: new Representation Autoencoders (RAEs) leverage frozen vision foundation models (VFMs) as tokenizer encoders, providing robust high-level representations that facilitate fast convergence and high-quality generation in latent diffusion models. However, freezing the VFM inherently constrains its spatial reconstruction capacity, limiting fine-grained generation and image editing; in contrast, incorporating reconstruction-oriented signals via fine-tuning disrupts the pretrained semantic space and degrades generative fidelity.