PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
April 30, 2026IEEE Transactions on Visualization and Computer Graphics0 citations

X2Video: A Novel Diffusion Model for Guiding Neural Video Rendering

X2Video: Adapting Diffusion Models for Multimodal Controllable Neural Video Rendering

View Full Paper
Ask AI
Bookmark
Share

Authors

ZHZhitong HuangMZMohan ZhangRWRenhan Wang

Discussion

Loading...

Member takes

Overview

Randomized trial demonstrates enhanced photorealistic video rendering with intrinsic guidance and multi-modal controls.

Key Points

  • The aim is to develop a diffusion model, X2Video, for rendering videos with control over various intrinsic factors.
  • Utilized Hybrid Self-Attention for maintaining temporal consistency in video frames.
  • Developed Masked Cross-Attention to effectively manage global and local prompts for video rendering.
  • Created a video dataset, InteriorVideo, consisting of 1,154 rooms to support the training of X2Video.
  • X2Video produced long, temporally consistent, photorealistic videos guided by intrinsic conditions.
  • The model supports multi-modal controls, allowing adjustments to color, material, geometry, and lighting.
  • Evaluations showed significant fidelity to reference images with effective parametric tuning.

Cite This Study

Huang et al. (2026) studied this question.

synapsesocial.com/papers/69f2f1471e5f7920c6386fe7https://doi.org/10.1109/tvcg.2026.3687740
View Full Paper
Ask AI
Bookmark
Share