 Hi everyone, so my name is Swan Martinez and today I'm going to talk about open source for the time collaborative workflow in Blender in one word sharing so I'm currently doing a PhD in partnership with CUP Creative and the in-rate research team from University Paris 8 my researches are focused on the link between real-time technologies and animation production such as how animation production could benefit from real-time in terms of communication tools So I will begin by explain some limitation of today's animation pipeline then I will show you an experimental solution and finally I will Confront it to both industrial and educational context Before starting I just wanted to say that There is other good existing solution to achieve real-time collaboration such as pocket studio and video universe or even game engine integration such as the multi-user plug-in for an engine St. Fusion for unity, but none of these were providing what I need such as open sources or a Blender compatibility because I needed to experiment collaboration into a multi-purpose DCC directly into it Let's give some context first Today's animation pipeline have been industrialized. What does it mean? It's mean that it rely on a linear assembly line A linear assembly line is a manufacturing process in which parts are added as the semi-finished product move from Workstation to workstation in animation workstation means production stages such as storyboarding lighting rendering and compositing and Parts could either be asset or shot So we have all those shots and asset line re-created through all those stages To bring the movie to life simple isn't it? But today's animation production looks more like that artists working on Individual tasks behind the screen There is no communication between artists from different stages those wall of isolation is Keeping them from knowing what happened before and after that tasks There's wall of silence could lead to mistakes Here for example the anime since the animator doesn't have any lighting information He couldn't prevent his character from casting in wanted shadow on the right bottom of the picture during the lighting state Those kind of could impact the wall projection chain and can be really tricky to roll back and cost a lot of money all those all those limitation From liner assembly line could came from a simple fact Between stages we're only sharing results not creation process So how to bring back communication between stages and Give back some project visibility to the artist Let's take some step back In IT Developers are using a giant methodology to collaborate easily on project and get a better project visibility In those methods an interesting concept is pair programming It consists bringing multiple developers to work on the same code base at the same time By doing this way they are sharing their knowledge and Made better decision to orient development so Why doesn't apply those concepts in animation and call it per creation for example Yeah, let's try that So in my opinion per creation Consists to bring multiple artists to work on the same on the same piece at the same time as synchronously but more precisely in animation Per creation consists to bring multiple digital artists from different or same stages to collaborate on the same scene at the same time but Putting multiple user into the same shared space in just some Problematics like how they will see each other's how they will communicate to To collaborate because the communication is the case for a solid collaboration and an efficient collaboration and Finally all to prevent conflict between users With all those problems in mind, I made the blender multi-user add-on which consists to bring Multiple artists into the same blend at the same time and left At the left you have a basic user representation with user selected objects and user first room and At the right you have all session related settings such as the user list reflected actions and replicated properties actions The add-on is based on a basic client server architecture with a data oriented application because Blender the doesn't allow me to Make operational based replication because of context conflict I also took some inspiration of git based mechanism to give some flexibility into the replication pipeline So we have a basic add-on which can put multi-use Multiple user into the same space and do things together. Let's try to let's try it in both industrial educational context But before starting I just want to specify that all industrial tests were made at cube creative with Tangranimo asset. Tangranimo is a cartoon TV show produced by CubeCrate Our first try consisted to bring Four CG supervisor in the following scene no rules given I just asked them to try to enrich the scene with the existing And we end up with that a total chaos Because without any direction and Without any direction and To soft right management user Users were interfering with each others and doing nothing productive. This is what I call the sandbox effect and Strong of this knowledge we set up a second experiment with three artists this time Tangi, Clemence and Tristan. I assigned a collection per artist to avoid conflict and They started with an empty scene and With the following asset pallets from Tangranimo to make a background concept Here you can see Clemence making some nice rocks while Tangi is adjusting the lighting color in real time Those kind of feedbacks are possible thanks to Eevee This time the session was more productive. This is the further the result and An interesting fact was the spatial user work repetition in yellow You have Tristan work in green you have Clemence work in red you have Tangi's work and at the beginning Each of them were working in his own corner, but as time passed They started to work together to bring the island to life by using oral communication as A way to organize themself They started to break the wall of isolation We learn another interesting fact Revote solid spatial guidelines the user doesn't know where to start Like here you have some wood without purpose Because they don't start with strong foundation So it could like it could lead the lead the artist to a lack of consistency in the background composition with that in mind We set up Thanks to Claire Nathan a third experiment only Tangi and Clemence this time Assets from Tangranimo still there and like before they put so its foundation first together so Here is the time-lapse of the experiments you can see Tangi in blue and Clemence in green Working together they started to put some heels then They are bringing the environment to life and it will come to a point where Tangi Will bring some nice rocks together while Clemence is addressing the The cloud and Something interesting Clemence just put Tangi into shadows So Tangi immediately told her to move his object away this kind of feedbacks and Responsiveness is given by the real-time aspect of the collaboration and Allows Tangi to react quickly on Clemence mistakes in a traditional linear pipeline Mistakes like that took more time and money and Here is the result of the third experiment more consistent this time because they started to put common foundations first I will rapidly talk about my Experimentation in teaching blender with the multi-user add-on so I let seven students work together on one common project they done lighting Compositing and a bit of shedding and modeling together on one shared space on one shared blend and It turns out that being in the same space allowed them to share feedback together and to to be more To go to to be more easily to help them together and It also allows me to help them with what quitting my chair very useful So this the add-on is far from finished now, but I learned some interesting fact from all those sessions such as the coordination effort is much More complex than in traditional linear workflow because without solid right management it it's users Without the solid right management, there are conflict between artists a So it foundation need also to be provided since the beginning because without any solid foundation a lack of consistency starting to show to be and but Oral communication become another natural way to share feedback and Allow user to react with each other in the natural way and lead them to better decision artistic call decision and Allow them to prevent mistake Some artists also told me that it's make the creation fun again until now the multi-user add-ons known some limitation like slow performance on complex maces and To to tackle with that I I easily used collection instances to let blender handle the eevee eevee data Until now not all data block types are supported yet like animation or Particle, but I'm working on it and Since we hadn't any need for new communication channel directly in blender because we don't alter all of our tests locally I Didn't develop like in a communication system like internal chat or VoIP What next? firstly, I will start by bringing multiple animators on the same scene So animation support will be a first step Then I will add a better support for differential revision in order to reduce network to reduce a network run with and Then I will start to Put the supervisor in veer in order to help him to get the bigger picture of his artist Working together Finally, I will earn some quality and time evaluation to To evaluate the mason And my guess is a future where a real-time Collaborative workflow will allow artists to work together on the same on the same movie and to bring all the proof to Allow together to work as an orchestra to bring the movie to life coordinated by supervisor and director Thank you all for listening