We propose a novel latent-semantic diffusion model that enhances image generation quality and training efficiency by integrating low-level latents and high-level features, simplifying training and improving inference.

https://arxiv.org/abs//2504.16064

YouTube: https://www.youtube.com/@ArxivPapers

TikTok: https://www.tiktok.com/@arxiv_papers

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers