OPEN LINK ↗
// 53d agoOPENSOURCE RELEASE
See-through automates anime layer decomposition
Shitagaki Lab's AI framework automates the tedious process of decomposing static anime illustrations into up to 23 semantically distinct, inpainted layers. Optimized for 2D animation tools like Live2D, it exports directly to PSD files, significantly reducing manual segmentation work for artists.
// ANALYSIS
See-through is a major efficiency boost for the VTuber and 2D animation industries, effectively automating the "cutting" phase of Live2D production.
- –Conditionally accepted to SIGGRAPH 2026, validating its technical novelty in stratified layer generation.
- –Unlike generic segmentation models, it "hallucinates" occluded regions (hair behind shoulders, etc.), which is the most time-consuming part of manual prep.
- –Low-VRAM mode (8GB) makes it accessible to independent artists and hobbyists, not just large studios.
- –Integration into ComfyUI shows rapid community adoption and desire for streamlined generative art workflows.
// TAGS
see-throughimage-genautomationopen-sourceresearchmultimodal
DISCOVERED
53d ago
2026-04-05
PUBLISHED
53d ago
2026-04-05
RELEVANCE
8/ 10
AUTHOR
AI Search