Animation, Graphics, and Imaging

Wolfgang Heidrich (leader), Pierre Poulin (co-leader)

The artistic and technological components of animation range from traditional techniques to leading edge digital tools used for creating animated content and special effects for film, video, games, and websites. One of the biggest challenges in animation and games is content creation. This encompasses static and animated geometric models of objects or scenes, appearance properties, physical properties, motion for virtual humans and other characters, and character behaviour and believability, to name a few. These are commonly designed by expert artists and programmers essentially "by hand" in an ad hoc process with tools that provide little more than raw capabilities. Because this is time and labor-intensive, it sets up barriers to scalability for companies and also to the possibility of amateurs engaging in creation.

Note, for example, that the development time and budget of a major game now is in the same range as the biggest films. In both games and film, the economics of this process frustrates the demand to go further. In contrast, the emergence of a Web 2.0 demonstrates the desire of non-experts to engage in content creation. As these users create, in essence, a culture of consumer-as-creator -- new media (such as video blogs) and creative/participatory games (such as Spore) become a major cultural force. In response, the field of computer graphics research has begun to shift from investigating raw capabilities towards more scalable methods to create content and build interactive worlds.

GRAND's AnImage theme will not only tackle the question of how to create content, but will also question what to create, including ways to intuitively guide creators toward the details that matter. This is not just a question of economic scalability, but furthers the impact of this research in terms of social foundations and quality of life by integrating interactions with other GRAND themes. Areas of focus include the following.

Animation

A prominent approach for generating animation in both games and movie special effects is to simulate the laws of physics in order to generate realistic results. This approach is suitable for generating a wide range of different effects, ranging from the simulation of natural phenomena (such as liquids, gases, and fire) to human (and other) characters, as well as passively deforming objects such as garments or plants. A central research topic is the analysis of human motions (MOTION) and understanding how these can be controlled and edited in physical simulation environments (CAPSIM). Another focus is the augmentation of simulators for fluids and other natural phenomena with models derived from real-world observations.

A complementary approach to physical simulation is evolving in the form of methods for effectively exploiting cameras and other sensors to acquire or measure real-world motions or performances. The goal is to extend traditional motion capture to include full characters, with garments and detailed characteristics such as facial models and hair, as well as to capture natural phenomena such as liquids interacting with each other and with solids. The captured data will then be analyzed to derive computational models for phenomena that are difficult to simulate (PROMO). The resulting models can then be integrated into simulation frameworks. This goal involves subfields such as computational photography (extending the capability of cameras to capture more data), vision-based model reconstruction, and learning compact-but-flexible models of human motion from captured motion and deformation data.

Scalable Content Generation

Another topic of this theme will be the development of algorithms that leverage computational power in order to extend the ability of a user to create compelling animation, character behaviours, geometric models, and so forth. A central technique for generating such scalable models is procedural modeling (PROMO). Understanding how to control procedural modeling and to generalize it to different scene contexts is key to the automatic creation of complex, detailed, and potentially infinite worlds. Moreover, by augmenting procedural modeling to estimate visual and gameplay impacts of scene elements not yet created, procedural modeling allows to handle scene elements on demand and at the appropriate level for any viewpoint.

Effective and Intelligent User Interfaces

The work on simulation and capture is complemented by research on novel user interfaces for manipulating or creating content. Of particular interest are new ideas in sketch-based modeling, interfaces using the sense of touch (SKETCH), and intelligent editing tools. Examples include tools that can automatically segment a model into semantically meaningful parts, so that an edit in one region can propagate to the rest of the model appropriately. Design of effective user interfaces is also crucial to provide user-control to procedural modeling, allowing some creative freedom in this traditionally one-way scene generation. These interfaces also consider using real data taken from photos, and using its extracted information to control content creation, or motion and deformation.

Human Perception

Finally, it is mandatory to understand the parts of a digital experience as they are perceived by humans in the context of different game genres or films. For example, the limitations of the human visual system must inform decisions of which effects to simulate, capture, or design interactively. This understanding is crucial for developing compelling animations, virtual worlds, and special effects that can communicate an artist's intent more effectively. It is also critical to the inference of meaningful semantics in geometry and motion. The perceptual research in this theme will therefore be tightly integrated with the development of novel user interfaces (AMBAID), and new capture and simulation techniques.

This theme will therefore not only result in new knowledge and technology, but also in new visions of creating content that is adaptable to the desires of a scene designer or animator. As such, providing mechanisms for effective feedback and for editing will be a fundamental aspect of our approach.