Thanks for reading our supplementary materials. 
We provide rich video results in the folder named 'videos_organized_by_anonymous_homepage'. Please use browsers to visit index.html file in this folder. All video files are local files, and we made this homepage only for better reading experience. No outside links are contained in this supp. mat. We include some reference images of dialogue videos in the folder 'input_imgs'. For each person, we reuse the same image in our experiment. For videos not showing reference images, you could assume the reference images is all person heads or their holding objects.

We also provide a code re-implemented upon OmniAvatar with Wan2.1 DiT for better understanding our method, as our original pretrained DiT could not be publicly available. Please refer to the folder named 'InterActHuman_code'.

