Gojo lab parses line drawings into joints, boundaries and correspondences
debreuil · x · 2026-10-24
Developer debreuil showcases Gojo · visual evidence laboratory, an interactive experiment that parses line drawings into inspectable evidence: object declarations, boundary walks, and contact correspondences, with color-coded reads (closed boundaries, incident directions at shared sites).
Key points:
- Encoding happens across many stages from rough outline to detail; different priors (organic vs man-made) interpolate/hallucinate differently.
- Only graph-verifiable evidence is encoded — geometry and connectivity are checked after save/restore.
- Explicitly excluded: human drawing order, hidden depth, ideal-shape recognition, or learned probability; imported icons keep path continuity while object identity and intent remain open.
A thoughtful demo on the boundary between visual evidence and hallucination in vision models.
Related event: Experiments explore structured visual evidence parsing of line sketches(2 posts)→
More from Multimodal
- Community fine-tune fixes Qwen-Image 2.1 VAE 8-pixel grid artifact — Crazy-Repeat-2006 · 2026-10-24
- Character LoRA trained on 21 images fails across models—training setup questioned — IntrovertedOtaku03 · 2026-10-24
- Tencent's Hy-Nomas video world model ported to native ComfyUI, runs on 12GB VRAM — LatentSpacer · 2026-10-24
- Video models replace NeRF/3DGS pipelines: single-frame novel view synthesis, open-source soon — jonstephens85 · 2026-10-24
- Recreating PS2-Era Harry Potter With FastH3-V2 in 4 Steps — aziib · 2026-10-24
- A Full AI Music Video Made Locally on DGX Spark With MiniMax H3 — Ornery-Block-3522 · 2026-10-24