A cutout alone cannot make a character pose: when hair or clothing moves, the image must contain pixels that were hidden in the original. This project tackles that harder step by reconstructing occluded regions as well as separating visible parts, turning one static illustration into a more useful 2.5D asset.
See-through: Single-image Layer Decomposition for Anime Characters
Turns a single anime illustration into up to 23 semantically organized, inpainted layers with inferred drawing order, exported as a layered PSD. It provides a 2.5D animation starting point, but does not handle Live2D rigging or motion.
Introduction
Information
- Websitegithub.com
- OrganizationsSaint Francis University, University of Pennsylvania, Spellbrush, Shitagaki Lab
- AuthorsJian Lin, Chengze Li, Haoyun Qin, Kwun Wang Chan, Yanghua Jin, Hanyuan Liu, Stephen Chun Wang Choy, Xueting Liu
- Published date2026/03/31
Categories
More Items
AI Video Papers2026
Generates subject-consistent videos from multiple reference images, using reconstructed surface-normal maps to guide geometry as the camera moves around the subject.
Conditions robot action generation on predicted visual and motion features, capturing physical dynamics without generating future video during control. Reports results across four simulation benchmarks.
AI Video Papers2026
Generates continuing video segments from multi-turn natural-language instructions and camera control. Its segment-wise causal attention preserves scene context while confining action instructions to their intended time windows.