- Also read the related article, which we made available to you free of charge on 26.01.2022
- We linked you to the complete short film “Cosmos Laundromat” on 27 January 2022 as part of our “Animago Afternoon” series
Mathieu Auvray is a French director who can look back on almost ten years of experience in filmmaking. He started out as a 2D artist at Nobrain, which later became Sabotage Studio. There he worked on music videos, TV idents and film credits. He slowly began to orientate himself towards 3D. Blender was his tool of choice.
He began realising entire short films in Blender for the company Autour de Minuit. He also created a TV series called “Babioles”, the pilot episode of which won an animago AWARD in 2010.
DP: Mathieu, how did you come across Blender?
Mathieu Auvray: I started with video editing and then became a 2D artist. I first worked as an editor on documentaries. As I had briefly attended art school, I immersed myself in the world of computer graphics. First as a matte painter, then as a designer for TV idents. I moved on to 2D animation and compositing for music videos, commercials and VFX. Everyone I worked with was using 3ds Max or Maya as their 3D software and it took me quite a while to get to grips with one of these programmes. But our Technical Director was a big fan of open source – and I liked Linux, Firefox and other open source software. So he showed me Blender and I thought to myself: “Well, since it’s going to take me several years to get to grips with any of these 3D software programmes anyway, why not start with open source?” It really took me several years to learn how to use Blender, and I’m still learning. But it became my favourite filmmaking program. And because you can really create complete films with it, I started making my own. That’s how “Babioles” was born. Here’s my very first test, which was six years ago: vimeo.com/2265583. I still had to do the tracking with PFTrack back then, because Blender didn’t have its own tracking module yet. When we really started with “Babioles”, we were already able to use the tracker in Blender a little. It was still at an early stage of development, but it was already extremely helpful.
DP: “Babioles” won an animago AWARD and a Suzanne Award from the Blender Foundation. Was that your foot in the door with “Project Gooseberry”?
Mathieu Auvray: The “Babioles” short film was intended as a pilot for a TV series. But the production company I work with in France wasn’t big enough to produce 24 episodes of such quality. We also had difficulties finding 3D artists with Blender experience in France. So we contacted Ton Roosendaal and went to the Blender Institute in Amsterdam to meet the team there and see if they could help us. Unfortunately, it turned out to be difficult to mix open source and creative commons with a closed, commercial project. So we went back to France and trained people ourselves and produced the whole series on our own. But we had help from Nozon, a post-production company. They helped us with rendering and compositing. The episodes were created entirely in Blender, with the exception of the compositing, for which Nuke was used. Since the visit to the Blender Institute, Ton, Nicolas (the head of Autour de Minuit) and I have stayed in touch. A few years later, Ton invited me to the Blender Conference. That’s when he introduced me to the idea of “Project Gooseberry” and offered me the chance to direct it.
DP: What was the pre-production like? What ideas did you discuss until you agreed on the concept “Sheep looking for a more interesting life”?
Mathieu Auvray: We sorted out some ideas quite quickly, which was due to the nature of the project, namely that twelve studios were making parts of the film. The idea of a sheep who meets a salesman who sells him a more exciting life in other universes or life forms came to me early on. It was one of my first ideas and Ton liked it. I first worked with a French screenwriter, Régis Jaulin, a really great writer who specialised in animated films. What he wrote was very good, complex and a great film – on a par with Pixar and Sony films in terms of the script and the challenges. But we couldn’t realise that. So I tried to simplify the aim of the film. I wanted it to be more of an independent film. I also wanted to work with Esther Wouda because I liked “Sintel” and felt that a female screenwriter could bring more sensitivity to the otherwise male-dominated animation world. So we met up and one stone fell on top of the other. We were quickly in tune with each other and what I am particularly pleased about is that the final plot is very close to what we had imagined in terms of emotions and feelings from the outset.
DP: How would the core team in Amsterdam have kept the threads together in the original concept? How much freedom would the external studios have been given in terms of style and narrative pace?
Mathieu Auvray: There was some debate about that. Each studio would have basically made its own film, and we would have only provided the script and some guidelines. For me, that was neither fair nor realistic. I would have been honoured as a director for the work of others. There would certainly have been problems with continuity. It just didn’t make much sense to me in terms of the direction of the speakers, the pace of the story, the emotions and so on. So we talked about it for a long time and came to the conclusion that we would at least provide storyboards. The art direction would have been completely free, but we would have given instructions on emotions and mood. We would have needed a good connection between the studios and the Blender Institute, almost 24 hours a day due to the different time zones.
DP: That would have been a real stress test for the planned shot management, where you could only build on the basic work of “Tears of Steel”. How extensively did you discuss technical details and goals during pre-production?
Mathieu Auvray: We discussed a lot. Each of us was excited and scared at the same time. You need a lot of experienced artists from experienced studios to avoid simple mistakes. In addition, each studio has its own recipes for how they manage a production. Finding a balance that suits each workflow, standardising things, would have been a lot of work. But Francesco, our TD and pipeline developer, has done a good job with asset management, distributed rendering and workflow. His design is interesting and needs to grow and be tested in a real production environment. The pilot episode has now fulfilled its mission of developing basic tools and testing them. Now the workflow can be trialled with several studios.
DP: To what extent were you involved in the crowdfunding campaign? From the outside, it seemed a little premature. Important features and information were rolled out when the campaign was already running.
Mathieu Auvray: I worked on the script, designs, ideas, gathered information about the teams and their graphic worlds. I was also working on a short film in France. But Ton and the Blender Institute put their heart and soul into the crowdfunding. That’s just Ton’s way of doing things: Not waiting until everything is ready, but getting started right away. That’s great, but it can also lead to mistakes and it can happen that some things are published prematurely. But at least you are doing something and can test in real time. One or two things could have been better prepared, but failure is still relative here. For a project without a big name, we still had a lot of success. And that was Ton’s part in the production. He supported us and gave us a free hand with the script, so I didn’t interfere with the crowdfunding either. That was his field, and he showed that he was successful there, even if it didn’t become the next internet hype.
DP: Did you already have a plan B in case the crowdfunding target wasn’t reached?
Mathieu Auvray: We had discussed all possible scenarios. My biggest concern was that the target would be reached. Then I would have had to give up my sleep for 18 months. The idea of the pilot episode was an option early on.
DP: Working with a small team directly in Amsterdam was certainly something completely different to what you had prepared for?
Mathieu Auvray: The team in Amsterdam was part of the plan from day one. Working there was going to happen anyway. But then I would have been mainly busy supervising the work of the external studios. And I wouldn’t have had the chance to travel around the world and get to know every studio personally.
DP: Is an open film very different from a conventional production? You had to work with software that was constantly evolving and in a kind of permanent alpha state. On the other hand, you had developers on site to implement features exactly according to your wishes.
Mathieu Auvray: I don’t think there are many small studios with so much in-house support. It was incredible to ask for a change to such a complex programme – and have it on the desk for testing a few minutes later. I didn’t find it uncomfortable at all. There were crashes and bugs, but overall the work was never blocked or slowed down as you can always go back to more stable versions. This is also one of the really good things about Blender. You can use any version, there are almost no conflicts if you know what you’re doing. Having the chance to show developers directly how to use their tools is fascinating. Sometimes you do it in a way they would never have dreamed of. You can discuss the best way to implement a feature based on your own experience. At the same time, you can make sure that it fits in with the existing code reactions and because you are working directly with it, stability is also guaranteed. I remember long discussions with the team about video editing, how editors use NLE software and why they do it this way. As a user, this is also interesting and challenging, as you can learn in this process how inefficiently you yourself use the developers’ tools. It shows you all the things you didn’t know about the software – and how lazy you can be.
DP: How many of the technical goals were set before the actual production started? For example, when did you decide to do without compositing and render everything directly in Cycles?
Mathieu Auvray: There were many technical goals for this project. These included asset management and the online rendering tools to make the feature-length film possible. Ton also wanted improvements in the area of physics simulation. So we opted for the sheep, the grass and the tornado – bam! There were also video editing challenges and the Dependency Graph refactor on top of that. As Cycles can create beautiful images without post-processing and we didn’t have much time for fine-tuning and compositing, we decided to render everything directly in Cycles. This saved us a lot of time and manual work, but the rendering times skyrocketed. However, it was helpful that we had many similar shots in the same environment. Eight minutes of the film take place in the same location in the same environment with only two lighting situations, namely early morning and the storm. Once you have a good lighting set-up that works in most of the shots, you only have to change very little to make them fit together. However, I also wanted the finished film to contain motion blur. But because the scenes were so big and had long render times, we had to do without it.
DP: How does the average Blender user benefit from the “Gooseberry” project or the “Cosmos Laundromat” pilot?
Mathieu Auvray: I don’t know how many bugs were discovered and corrected during production. I also can’t say anything specific about the many improvements to features and performance, but it’s enormous! Also, the refactorings are sometimes not directly visible in the UI. A major overhaul of the code can take place in the background and then open up new possibilities. Sergey and Lukas worked a lot to clean up code and document it for future refactorings. This is invisible work, but it is necessary and takes a lot of time. With many large software packages, this happens far too rarely. They simply add features and make sure that they work. Here at the Blender Institute, we are constantly trying to optimise things and make sure that they are programmed in such a way that future developers can also understand the code and add their own.
DP: What’s next for the “Cosmos Laundromat”? Do we have a chance to finally meet Tara?
Mathieu Auvray: I hope so. That’s the most important part: the love story. It’s completely missing in the pilot. The Blender Institute has some smaller projects going on that will keep them busy until the end of the year. I know that Ton will soon make a decision about the future of “Cosmos Laundromat”. It depends on the feedback, especially since major studios and a major production company are sharing their thoughts on it with him.
There are three options:
1. We leave it at that and move on.
2. We succeed in financing the entire film.
3. We produce further episodes, initially seven minutes, with the income from the Blender Cloud.
DP: What was the reaction to the pilot episode? Both Dreamworks and Pixar invited you to a special screening in front of their employees.
Mathieu Auvray: I was on holiday in Thailand with my girlfriend at the time, so unfortunately I couldn’t go to Pixar and Dreamworks. But the team told me about their experience. There was great feedback. We received a lot of congratulations on the technical side, but the film itself was also enthusiastically received. At least there was enough interest for viewers to want to know what happens next. I also received a lot of positive feedback from people who watched the film online, but also from colleagues. They all had one thing in common: what happens next? That was our main goal, as the production had been so heavily condensed. The tonality of the film and the feelings it evokes touched many people – and also surprised them, as “Cosmos Laundromat” is rather unusual for an animated film. I am very proud of the fact that most people want to know how the story unfolds. What’s more, they hardly notice that the film is 10 minutes long. That’s really fantastic, because that’s exactly what we wanted to achieve without having to resort to a classic cliffhanger.
