TOOLS // ANIMATION | BUILD
Based upon its expertise in the
field of facial feature tracking and expression recognition, Dynamixyz has developed a statistical tool making it possible to replace motion capture.
Gaspard Breton, Dynamixyz
What is Dynamixyz’s experience in the games industry? Originally, the founders of Dynamixyz come from a different technical world, where markerless, video-based tracking techniques are the state-of-the-art. Now, this technological framework encounters a growing interest in the graphics community, as it helps solving problems that were not well addressed by existing hardware and software; typically in facial animation. The knowledge and expertise of
Dynamixyz is thus very valuable in the establishment and democratisation of those new technologies. The work of Dynamixyz is now fully
committed to the world of 3D animation, and especially to the entertainment industry, meaning films and games. Among our customers and partners we count Motek, Animatrik and Orange.
You’ve got a new version of Performer coming out very soon. What is that tool, and how have you pushed it forward? Performer is Dynamixyz’s software solution for video-based markerless facial tracking and motion retargeting. It is constructed as a stand-alone software that can be installed directly in studios, and allow operators and animators to process everything themselves, and in-house. It can process video from existing video capture tools, and interfaces with standard animation tools such as Motion Builder or Maya. The output of the system for an input video is the corresponding animation curves on any type of character rig. The Performer software has existed for some time, and is now being used in production. The technical core has proved its accuracy and robustness in facial tracking, so for the latest versions we’ve focused on refining its performance, particularly its processing speed. Another new available technical feature is the motion transfer of eye movements. The biggest improvement over the last
versions, however, is probably how Performer interfaces with existing production pipelines. More standard outputs have been added, such as the C3D format for tracked facial points. We’ve also worked on the support of rig specialisations such as driven-keys.
on
to achieve good quality. It was important to us to be able to offer a complete solution, which is why the Dynamixyz tools cover the whole chain of facial animation production. Each of the modules of that chain, however, has open data formats, and can be used independently to better integrate to different production pipelines. On a more technical aspect, what defines
the technology is that it uses markerless tracking, which is very well adapted to the characteristics of facial motion capture.
DEVELOP-ONLINE.NET
How will Performer directly impact developers and the games they make? Only a small portion of studios have the personnel and equipment to produce performance-based facial animation, let alone at large scale. Performer looks to address that as the requirements in equipment and animation expertise are lower than for other existing solutions. Facilitating the process of creating
performance-based facial animation enables studios of all sizes to produce at competitive costs. This means higher animation quality, and more facial animation in games. The higher- level impact is in a way the ability to create more realistic, compelling virtual characters.
What challenges do you see as key for the facial animation sector to overcome? As many actors of the games and movie industries relate, facial animation is the next
big step. Now this is two-fold; there is obviously the challenge of quality, which so far only a few studios can guarantee, and there is also a challenge of accessibility, meaning that every studio should be able to produce high-quality facial animation. Better accessibility will mean more quality results in movies and games, for the benefit of players and spectators. Dynamixyz looks to address both challenges. The tools we provide are quality-driven, but they are also meant to combine naturally with traditional pipelines and to be accessible to non-experts. What lies ahead now is the integration of disparate tools – for facial and body motion capture typically – to achieve the optimal conditions for performance-driven animation. This will clearly mark a step forward in the production of quality animation and making the most out of performance-driven characters.
What is needed to tackle those challenges? Could collaboration be a solution? Collaboration is definitely the key. As powerful facial animation tools can now be provided by markerless and video-based techniques, whose use is pretty recent in the game industry, there’s sort of an adaptation phase, where the integration of those techniques and the more traditional ones needs to be defined and optimised. We are already working on that process with several studios, and we’re really glad to be part of those types of collaborations. It really feels like combining complementary skills and knowledge, and creating a synergy achieve the best possible animation quality.
Dynamixyz has stated previously a belief that there is an immediate need to develop statistical systems which can produce good quality, on-the-fly animations. What does that mean? Indeed, based upon its expertise in the field of facial feature tracking and expression recognition, Dynamixyz has developed a statistical tool making it possible to replace motion capture. The concept behind it is to learn from a
large amount of facial tracking data the facial behaviour of a particular actor in order to be able to produce on-the-fly realistic animations. We call this, an ‘expressive space’. It is very straightforward to navigate in this kind of area. To make up ones mind, you can imagine a plane travelling from one waypoint to another, each being a trend in the facial expressivity. We believe this kind of tool can be a really interesting alternative to use for the video games industry because some games, like MMOs, have a lot of characters, each needing quite a large amount of animation data. Using middleware like the one we’ve
developed makes it possible to control all these characters realistically and in realtime using only scripts. Moreover, expressive spaces offer a larger scale of expressions, and prevent animations from looking repetitive, adding randomness to the final output.
www.dynamixyz.com
AUGUST 2012 | 73
Dynamixyz CEO Gaspard Breton is confident his company’s experience in other industries can benefit games makers
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76 |
Page 77 |
Page 78 |
Page 79 |
Page 80 |
Page 81 |
Page 82 |
Page 83 |
Page 84 |
Page 85 |
Page 86 |
Page 87 |
Page 88 |
Page 89 |
Page 90 |
Page 91 |
Page 92 |
Page 93 |
Page 94 |
Page 95 |
Page 96 |
Page 97 |
Page 98 |
Page 99 |
Page 100