K² · Artificial intelligence
To texture 3D objects, film them with an AI video camera first
The authors introduce Ink3D, a framework that bridges 3D generation with large-scale video generative models to synthesize extremely complex textures.
YH
CL
Yue Han, Chong Li, Zhening Liu et al.
8 authors · cs.CV
The 30-second scan
AI can already sculpt a decent 3D shape from a picture.
- Ink3D first reconstructs a white-mesh geometry using an off-the-shelf 3D generation model.
- Ink3D employs OrbitPainter, a conditional video generative model, to produce dense orbit-scan videos capturing object appearance across viewpoints.
- The authors introduce TextureOptimizer, a neural baking module that integrates dense multi-view observations while mitigating geometry inconsistencies arising from video generation.
Visual generative models are trained on datasets several orders of magnitude larger than 3D datasets