8 papers
Physics in 2-Steps: Locking Motion Priors Before Visual Refinement Erases Them
Woojung Han, Seil Kang, Youngjun Jun +3
Image-to-Video diffusion models leverage input images to generate visually stunning content, yet frequently produce motion that violates physical laws. We reveal a surprising findi…
Real-Time Visual Attribution Streaming in Thinking Model
Seil Kang, Woojung Han, Junhyeok Kim +3
We present an amortized framework for real-time visual attribution streaming in multimodal thinking models. When these models generate code from a screenshot or solve math problems…
ViKey: Enhancing Temporal Understanding in Videos via Visual Prompting
Yeonkyung Lee, Dayun Ju, Youngmin Kim +2
Recent advancements in Video Large Language Models (VideoLLMs) have enabled strong performance across diverse multimodal video tasks. To reduce the high computational cost of proce…
Interpretable Motion-Attentive Maps: Spatio-Temporally Localizing Concepts in Video Diffusion Transformers
Youngjun Jun, Seil Kang, Woojung Han +1
Video Diffusion Transformers (DiTs) have been synthesizing high-quality video with high fidelity from given text descriptions involving motion. However, understanding how Video DiT…
Rare Text Semantics Were Always There in Your Diffusion Transformer
Seil Kang, Woojung Han, Dayun Ju +1
Starting from flow- and diffusion-based transformers, Multi-modal Diffusion Transformers (MM-DiTs) have reshaped text-to-vision generation, gaining acclaim for exceptional visual f…
Interpreting Attention Heads for Image-to-Text Information Flow in Large Vision-Language Models
Jinyeong Kim, Seil Kang, Jiwoo Park +2
Large Vision-Language Models (LVLMs) answer visual questions by transferring information from images to text through a series of attention heads. While this image-to-text informati…