PeCA: Palette Context Assisted Inference for Test-Time Paint-Bucket Colourisation on Animation Videos

2026/08/02 by Dongheng Lin, Jianbo Jiao
Computer Science · #cs.CV

paper · pdf

ECCV 2026, Project Page: https://rathgrith.github.io/PeCA/

arxiv created 2026/08/02 · arxiv updated 2026/08/04

Abstract

In animation production, paint-bucket colourisation for hand-drawn animation is a labour-intensive procedure that assigns each enclosed region in line sketches a colour from reference design sheets. Recent automatic paint-bucket colourisation pipelines mirror this workflow via region correspondence, but correspondences can be brittle when regions are ambiguous fragments without proper context. In this paper, we propose Palette Context Assisted (PeCA), a new training-free, plug-and-play framework for animation video colourisation that aims to close this gap at test-time via reasoning over spatial and temporal contexts. Extensive experiments on existing benchmarks and a newly introduced long-video test case show consistent performance boosts.

Citations

• V-Thinker: Interactive Thinking with Images
• DACoN: DINO for Anime Paint Bucket Colorization with Any Number of Reference Images
• ToonComposer: Streamlining Cartoon Production with Generative Post-Keyframing
• AnimeColor: Reference-based Animation Colorization with Diffusion Transformers
• Cobra: Efficient Line Art COlorization with BRoAder References
• Follow-Your-Color: Multi-Instance Sketch Colorization
• SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features
• LayerAnimate: Layer-level Control for Animation
• MangaNinja: Line Art Colorization with Precise Reference Following
• Generative AI for Cel-Animation: A Survey
• AniDoc: Animation Creation Made Easier
• ColorFlow: Retrieval-Augmented Image Sequence Colorization
• Paint Bucket Colorization Using Anime Character Color Design Sheets
• LVCD: Reference-based Lineart Video Colorization with Diffusion Models
• SAM 2: Segment Anything in Images and Videos
• ToonCrafter: Generative Cartoon Interpolation
• Generative AI for 2D Character Animation
• Sakuga-42M Dataset: Scaling Up Cartoon Research
• Learning Inclusion Matching for Animation Paint Bucket Colorization
• CharacterGen: Efficient 3D Character Generation from Single Images with Multi-View Pose Canonicalization
• Region-Based Representations Revisited
• MMICL: Empowering Vision-language Model with Multi-Modal In-Context Learning
• CoTracker: It is Better to Track Together
• Improving Video Colorization by Test-Time Tuning
• Emergent Correspondence from Image Diffusion
• DINOv2: Learning Robust Visual Features without Supervision
• Micrograph segmentations for DDEVD
• Segment Anything
• Zero-1-to-3: Zero-shot One Image to 3D Object
• Scalable Diffusion Models with Transformers
• Test-Time Prompt Tuning for Zero-Shot Generalization in Vision-Language Models
• High-Resolution Image Synthesis with Latent Diffusion Models
• High-Resolution Image Synthesis with Latent Diffusion Models
• Tip-Adapter: Training-free CLIP-Adapter for Better Vision-Language Modeling
• The Animation Transformer: Visual Correspondence via Segment Matching
• Learning Transferable Visual Models From Natural Language Supervision
• RAFT: Recurrent All-Pairs Field Transforms for Optical Flow
• RAFT: Recurrent All-Pairs Field Transforms for Optical Flow
• Deep Learning for Free-Hand Sketch: A Survey
• Learning Correspondence from the Cycle-Consistency of Time
• Efficient semantic image segmentation with superpixel pooling
• Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
• Distinctive Image Features from Scale-Invariant Keypoints