ToonCrafter

ToonCrafter is a research-driven project focused on cartoon interpolation: the process of filling the gap between two drawn frames so that motion looks smooth. Presented as an academic showcase rather than a commercial product, it lets animators and researchers test how generative models handle the distinctive look of hand-drawn and stylized artwork.

What it does

Given two input frames, ToonCrafter synthesizes intermediate cartoon images that sit between them, producing a short animated transition. It also supports sketch-based workflows, such as coloring a sketch using one or two reference images, and generating intermediate sketches when only sparse sketch guidance is avAIlable for the surrounding frames.

Key capabilities

• Cartoon interpolation: Creates plausible in-between frames from a pair of keyframes.
• Reference-based coloring: APPlies the palette and style of a reference image to a sketch.
• Sparse sketch guidance: Accepts partial sketch Information for the middle frames and completes the transition.
• Ablation studies: The project documents how indiVidual components affect ouTPUt quality, which helps researchers underStand the model.
• style consistency: Aims to keep line work and color treatment coherent across generated frames.

Who it is for

The tool suits animation students, illustrators and technical artists who want to experiment with assisted in-betweening, as well as researchers studying generative models for stylized imagery. Studios may find it useful for rough previsualization, but it is not positioned as a production pipeline replacement.

How to get started

Access is typically through the project page, which hosts dEMOnstrations, example results and technical documentation. Because this is an academic release, the exACT way to try it — an online demo, Sample gallery or code repository — may vary over time, and availability can change without notice. Review the documentation before relying on any specific workflow.

Pricing and access

There is no advertised commercial pricing in the source material, and the project appears to be freely viewable as a research showcase. Any hosted demo, compute requirements or usage limits may differ from expectations, so confirm current terms on the official page. The interface language is primarily English.

Limitations to keep in mind

The developers acknowledge challenges in understanding image content and in generating convincing transitional motion, especially with complex or ambiguous drawings. Results can degrade when the two keyframes differ greatly in pose or composition. Treat generated frames as a starting point that still needs an artist's review.