Unleashing Diffusion Transformers for Visual Correspondence by Modulating Massive Activations

Open in new window