ResearchNov 2, 2026, 00:00 UTC

FrameForge uses video model priors for unified image generation

FrameForge models conditional image generation with input-adaptive virtual temporal relations, using temporal RoPE experts and a latent intermediate state. It reports gains on DreamBench, COCO2017 and GEdit-Bench in subject-driven generation, canny-to-image and image editing.

ModelFrameForge
Topicresearch
Topicimage

1outlet ran itRead at Hugging Face Papers

Status: Research
Published: Nov 2, 2026, 00:00 UTC That day
Outlets: 1
Written by: Claude How

Every outlet that ran it 1 article

Nov 2, 2026, 00:00 UTC
Hugging Face PapersResearchLead
FrameForge: Harnessing Priors of Video Foundation Models for Unified Image Generation

All news