概要 About
Animation is no longer only hand-drawn anime. Models can now animate a still photo or synthesize stylized sequences — yet we still lack shared answers: How do we assess natural movement versus generated movement? What benchmarks capture temporal coherence, identity, and style? How do we control motion and animation with diffusion models and world models?
AINIME is, to our knowledge, the first workshop on this dual agenda: traditional anime/manga and natural-video animation. Creators need the tools; researchers need the venue. We bring generation, post-production, restoration, and quality evaluation onto one WACV stage — with a flagship Anime Restoration Challenge on Re-Anime600.
Built on the Re-Anime600 dataset and the AIGENS workshop at CVPR 2026.
募集 Call for Papers
We cover the whole production pipeline — from generation and post-processing to restoration and quality evaluation of natural and anime-style videos.
- Image-/text-to-video animation — natural and stylized
- Motion control — diffusion models and world models
- Natural movement vs. animation — new metrics and studies
- Benchmarks & datasets — temporal quality, identity, style
- Anime post-production — restoration, upscaling, colorization
- Responsible creative AI — authorship, consent, production tools
Regular papers, 5–8 pages (excl. references), double-blind review via OpenReview. The track is archival — accepted papers appear in the WACV 2027 workshop proceedings / IEEE Xplore.
| Submission deadline | October 20, 2026 (AoE) |
|---|---|
| Notification | October 30, 2026 |
| Camera-ready | November 20, 2026 |
| Workshop | January 4–5, 2027 · WACV Disney Springs |
Dates follow WACV workshop guidelines.
挑戦 Anime Restoration Challenge
The AINIME Anime Restoration Challenge is now open — the flagship challenge of the AINIME workshop at WACV 2027 (January 4–5, 2027). Push anime video restoration on a curated benchmark with real prizes and archival publication for top teams.
The task
Enhance the perceptual quality of anime video clips from Re-Anime600 while preserving the content, style, and temporal consistency of the original animation.
Re-Anime600 is a curated benchmark of 600 clips with high native quality — built for rigorous NR-IQA and super-resolution evaluation, not generic web video.
- 350 train clips — public after signing the data agreement
- 100 validation clips
- 150 test clips — hidden for leaderboard evaluation
Re-Anime600 — clips spanning low, median, and high quality across four NR-IQA axes.
Notice. We do not own the copyright of all content in this dataset. Re-Anime600 is released for research purposes only. Copyright holders may contact us to request removal of specific material.
Get started
Register your team, accept the data agreement, then download train data and prepare submissions.
Train data ↓
- Challenge hub on Hugging Face — leaderboard, rules, and updates
- Submission example (GitHub) — format and packaging reference
- Train access requires signing the AINIME Challenge Data Agreement on Hugging Face.
Important dates
| Oct 2, 2026 | Challenge opens · train data available |
|---|---|
| Oct 2 – Oct 20 | Validation phase |
| Oct 20 (AoE) | Deadline for challenge paper submission |
| Oct 21 – Oct 25 | Test phase |
| Oct 28, 2026 | Results announced |
| Jan 4–5, 2027 | Results & challenge session at AINIME · WACV Disney Springs |
Prizes
- Special issue publication: top teams will be invited to publish in a special issue of a Q1 Computer Science journal (regular peer review).
- Workshop registration: the AINIME workshop participation fee at WACV 2027 will be covered for one representative per winning team.
Challenge papers
Participating teams are invited to describe their method in a regular paper (5–8 pages, excl. references), double-blind via OpenReview. The track is archival (WACV 2027 workshop proceedings / IEEE Xplore). Paper deadline: October 20, 2026 (AoE).
Challenge contact: georgia.fargetta@unict.it · 1000069335@studium.unict.it
日程 Tentative Schedule
Half-day · times relative to the slot assigned by WACV (tentative morning).
| 09:00 – 09:10 | Welcome & overview |
| 09:10 – 09:40 | Invited talk — academia (animation / media / evaluation) |
| 09:40 – 10:10 | Invited talk — industry (production AI / localization) |
| 10:10 – 10:30 | Coffee & posters |
| 10:30 – 11:00 | Anime Restoration Challenge — results & baselines |
| 11:00 – 11:35 | Contributed orals / spotlights |
| 11:35 – 12:00 | Panel (metrics, motion control, data/IP, creator tools) & closing |
主催 Organizers
Primary contact: marcos.conde@uni-wuerzburg.de · georgia.fargetta@unict.it