← back to paper
arxiv: 2604.06757 · 2 revisions
FlowInOne:Unifying Multimodal Generation as Image-in, Image-out Flow Matching