|
|
|
| e-Pub |
Previous | 
Home
Section: Research Program
Validation methodology
When developing digital creation tools, validation is a major challenge. Researchers working on ground-truth reconstruction can apply standard methodologies to validate their techniques, such as starting by testing the method on a representative series of toy models, for which the model to reconstruct is already known. In contrast, it is not obvious how to prove that a given tool for content creation brings a new contribution. Our strategy to tackle the problem is threefold:
-
Most of our contributions will address the design of new models and algorithms for geometry and animation. Validating them will be done, as usual in Computer Graphics, by showing for instance that our method solves a problem never solved before, that the model is more general, or the computations more efficient, than using previous methods.
-
Interaction for interactive content creation & editing will rely as much as possible on preliminary user studies telling us about user expectations, and on interaction paradigms and design principles already identified and validated by the HCI community. When necessary, we intend to develop as well new interaction paradigms and devices (such as the hand-navigator we are currently experimenting) and validate them through user studies. All this interaction design work will be done in collaboration with the HCI community. We already set up a long term partnership with the IIHM group from LIG in Grenoble, through the INTUACTIVE project at Grenoble INP (2011-2014) which involves co-advised students, and through the co-direction of the action “Authoring Augmented Reality” of the larger Labex PERSYVAL project (2012 – 2020).
-
Lastly, working on specific applications in the domains we listed in Section 3 is essential for validation since it will give us some test beds for real-size applications. The expert users involved will be able to validate the use of our new design framework compared to their usual pipeline, both in terms of increased efficiency, and of satisfaction with new functionalities and final result. In addition to our work with scientific and industrial partners, we are establishing collaborations with the Ecole Nationale Supérieure des Arts Décoratifs (ENSAD Paris, Prof Pierre Hénon) and with the Ecole Nationale Supérieure Louis Lumière (Prof. Pascal Martin) for the evaluation of our ongoing work in shape and motion design, and on virtual cinematography.