OmniShowcase combines an automated data synthesis pipeline with geometry-aware conditioning.
Multi-view geometry conditioning preserves subject identity across 360-degree video generation
Ahn et al. train a video diffusion model on 4,000 synthetic renderings and reconstruct surface-normal turntables from reference images, outperforming prior methods on identity and 3D consistency.
Top University
Junyeong Ahn · Geonwoo Kim · Kinam Kim · Gihwi Kim · Hoonjin Jung · Hyojin Jang · +1 more
FLIPTION · KAIST AI · Inha University
Research Digest··2 min read
Ahn et al. introduce OmniShowcase, a subject-to-video generation framework that trains on synthetic 360-degree renders and conditions the video model on reconstructed surface-normal maps from multi-view reference images.
The Daily Edition
The front page in your inbox each weekday morning: the lead, the picks, and why each one earned its place.