Phd student at CUBE Creative and INREV (University Paris 8), my work is focused on what real-time rendering could bring on the table for computer animation in terms of innovative creation workflow and new narrative possibilities. During this first year my researches were focused on collaborative and immersive real-time workflow exploration.

A film is an idea carved along the whole production process by many different peoples. A traditional animation pipeline involve a linear succession of tasks. From storyboard to compositing by passing upon different step, its fundamental work flow is similar to an industrial assembly line. Since each step is almost a department, its common that one person on department B doesn't know what another person did on a previous step in a department A. This lack of visibility/communication could be a source of problems which could produce a bad impact on the final production result.

Nowadays it's a known fact that real-time rendering technologies allows to speedup traditional linear production by reducing drastically the iteration time across different steps. All majors industrial CG solutions are moving toward real-time horizons to bring innovative interactive workflows. But this is a microscopic, per-task/solution vision of real-time rendering benefits for the animation production. What if we step-back, get a macroscopic picture of an animation movie pipeline and ask ourself how real-time could change our global workflow ? Could-it bring better ways of working together by giving more visibility between departments during the whole production ?

Data flow and work flow are closely linked together since the data flow exist to support the work flow. Actual data flow standardization (USD, Materialx ,etc.) brings solids foundations for efficient inter-solutions communications, allowing studios to be more flexible on their technological choices. But these formats are not yet well-supported in open source solutions. Recently, my researches leads me to question if real-time interactions between different artists could create new dynamics. Blender support almost the whole film production steps, from storyboard to compositing.

In order to experiment a real-time parallelisation between different production stage I developed a pure python blender addon to bring multiple users into the same blender scene. By replicating blender data blocks over the networks, it allows different artists to collaborate on a same scene in real-time.

As I have the chance to do my thesis in both university and industrial environment, I can experiment collaborative workflows in both of them. At the time of writing those lines the multi-user addon is being tested in both. Firstly, I used it for educational purposes to teach blender modeling basics to university students. Then I made a first collaborative creation workshop with the same students in order to crash-test the addon and made a first study about frictions induced by real-time workflows parallelization (in this case modeling and layout of a landscape). The next step for August / September will consist to bring the multi-user addon into an industrial context at CUBE Creative. Different kinds of collaborative experiments will be conducted such as interactive review process, interactive realization assistance, tasks parallelization.

On one hand, this talk aims to present the result of this work. The result consists of showing our experimentations findings and the multi-user addon in action. On the other hand I'll talk about how I faced technical challenges specific to the real-time collaborative context (such as right management, user representation, etc.). Then I will explain more generally my guesses about collaborative workflow in the near future.