Motion Stylization for 2D Animation – MoStyle
Developing the stylized animation tools of tomorrow.
The central idea of the project is to bridge two very different forms of animation: 2D hand-drawn animation and 3D animation. In many modern productions, these two forms coexist. This project aims to help animators specifically with the design and stylization of movement, while giving them complete artistic control over their creations.
Helping artists to digitally create the typical effects of traditional 2D animations.
A growing number of animated series (e.g., "Arcane") and films (e.g., "I Lost My Body") are using hybrid production pipelines that blend 2D and 3D techniques, for example, animating characters in 2D and objects or sets (vehicles, buildings, etc.) in 3D. A sign of the industry's interest in this hybrid approach, the leading open-source 3D animation software Blender introduced a 2D animation environment within a 3D scene in 2019 with the "Grease Pencil." However, the animations resulting from this pipeline can be only moderately convincing, as the 3D elements tend to stand out—particularly in their movement—from the 2D animations, unless numerous iterations between 2D and 3D are performed to adjust each. Furthermore, this adds significant technical complexity, which already accounts for nearly 10% of the production costs of animated films. The main objective of this project is to study how computer tools can help capture and reproduce the typical characteristics of traditional 2D animation. More specifically, it aims to address the following scientific and technical challenges: 1) How can intermediate drawings (infill) be generated interactively and semi-automatically from a sequence of sketches to effectively define the movement of a 2D animation? 2) How can the geometry and movement of a 3D animation be stylized to facilitate its integration with a 2D animation? These two objectives target tasks that are currently tedious for artists. These tasks rely primarily on manual techniques or ad hoc solutions that severely limit artistic creativity and significantly increase production time. The solutions we will develop within this project must be compatible with standard animation workflows and provide artists with appropriate controls, placing them at the heart of the human-computer interaction.
In 2D animation, the starting point was the traditional way animators design character movements: a lead animator draws the most important key poses, then an inbetweener produces the numerous intermediate drawings between poses to give the illusion of continuous movement (12 to 24 drawings per second of animation). The succession of these drawings must create an impression of volume, so that the viewer imagines a "real" character moving and turning, and not just a simple sketch on a sheet of paper. The tools developed in this project aim to assist the artist from the rough drawing stage. Thus, even before an inbetweener creates all the intermediate drawings, the lead animator can see what the movement will look like. In other words, the computer program automatically generates a preview of the movement between two key poses. This is primarily a preliminary working tool to aid in the motion design of an animated film. Of course, this tool will not produce the final result; the lead animator can use this motion preview to share their work with collaborators (art director, producer, inbetweener) before starting the actual production.
In 3D, the challenge is somewhat different, as it involves helping to reproduce typical 2D animation effects with computer-generated imagery. While empirical techniques already exist, each animator has their own, and there is no general theorization of the processes used. Once again, the starting point was observing these techniques and discussing them with 3D animators, specifically in the case of stylized motion blur. To simplify the perception of these rapid movements, 3D animators who want to give their animation a 2D style will use various smearing effects, grouped into three categories: deforming an object in the direction of its movement; adding motion lines as in comics; and repeating a drawing several times and layering them. The strength of our project lies in unifying these three techniques and making them available in a plugin for the Blender software, which applies them directly to 3D animations. This simplifies the work of animators, as they can design and, more importantly, control these effects as they see fit.
To simplify and accelerate the creation of animated films, we have developed a 2D animation system that allows for motion preview right from the rough sketch stage. Using a set of semi-automatic tools, an artist can break down their sketched keyframes into motion units, whose trajectories, dynamics, and occlusions they can precisely control. The software then automatically generates all the intermediate frames between the keyframes in the artist's style, who can still interactively refine the animation's movement. This system has been described in three scientific publications and presented jointly with the company Praxinos at the "Animation Development Innovation Meetings" in Angoulême in 2024. Praxinos is developing 2D animation software within a 3D engine and has begun integrating some of the tools we have proposed.
For 3D animation, we have proposed a method for automating three traditional 2D motion effects for 3D animations of rigid objects and skeleton-animated characters, while offering extensive artistic control. This method has been presented in two scientific articles and is already integrated into a Blender extension that has been downloaded nearly 24,000 times to date (https://extensions.blender.org/add-ons/smear/).
Our work is also summarized in this short video by Curieux!:
While the 2D and 3D branches share a common goal, they follow different methods and development paths. The 3D component is already integrated into a plugin and available for download (https://extensions.blender.org/add-ons/smear/). It has already been downloaded nearly 24,000 times. The 2D component, on the other hand, remains more experimental. However, Praxinos, a company developing a 2D animation software within a 3D engine, has begun integrating some of the tools we have proposed. The next challenge: to further integrate these two branches to explore all the variations of movement in different types of animation (characters, perspective effects, special effects).
This project is positioned within the general context of computer graphics, focusing on the creation of animated films. Its main goal is to investigate how computer tools can help capturing and reproducing the typicality of traditional 2D animations. More precisely, it plans to address the following challenges: 1) How to interactively generate in-betweening frames from a sparse sequence of rough drawings to swiftly explore different animation choices and motion designs. 2) How to stylize the geometry and motion of a 3D animation to allow its seamless integration into a 2D animation. In both cases, the proposed solutions should be compatible with regular animation workflows and provide adequate controls to artists that should remain at the center of the computer-human interaction loop. Ultimately, those solutions will be integrated inside the same software framework, allowing to produce 2D animations with a unified appearance starting from roughs drawings or 3D inputs.
Project coordination
Pierre Bénard (Centre de Recherche Inria Bordeaux - Sud-Ouest)
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
Partnership
Inria Bordeaux Sud-Ouest Centre de Recherche Inria Bordeaux - Sud-Ouest
Help of the ANR 270,000 euros
Beginning and duration of the scientific project:
December 2020
- 48 Months
