Interview with Ton Roosendaal about Elephants Dream and free content movies

Sunday, May 21, 2006

Three days after the Internet release of the free content 3D short Elephants Dream (see Wikinews coverage), we exchanged e-mails with Ton Roosendaal about the reaction to the film, open source filmmaking, and the changes to Blender that resulted from the production. Ton Roosendaal is the lead developer of the Blender 3D rendering and modelling software that was used for the movie. He is also the chairman of the Blender Foundation, a non-profit organization which was formed in support of the software and projects like Elephants Dream.

How much money did the Blender Foundation spend on producing the movie? Has the money been fully recouped by DVD orders and donations?

We still have to finish the final bookkeeping for this project. It has been executed in co-production with the Netherlands Media Art Institute, and we each had our own internal budgeting for the project. When you exclude expenses of pre-production and producer personnel, the total budget was about 120,000 €, of which we covered half. Our contribution was roughly covered half by the DVD sales, and half by European Union support (http://www.uni-verse.org consortium).

One of the most common criticisms of CGI films is focus on technology over content. For instance, Final Fantasy: The Spirits Within flopped with audiences, in spite of being an undisputed technical milestone. I’ve seen many reviews that criticized the plot of “Elephants Dream” as too bizarre or confusing. In retrospect, are you happy with the story development process?

Haha, I knew the story and plot would get a mixed acclaim. There’s a couple of reasons I’d like to mention for it.

First of all; the criticism resembles how people witness Blender itself, too. Many people expect that Free Software is an easy accessible mass audience product. We get a lot of complaints by non-artists that they can’t get into the software easily, whilst the complexity of commercial products like Maya or Houdini is perceived as a confirmation of its “quality”. Apparently an Open Movie created similar expectations with the audience.

Luckily we also got many positive reviews of the artistic result of the movie. It is quite abstract, but definitely has many layers of information, inspiring many of the viewers to see relevant real life messages hidden here.

For this project we’ve teamed up with the Netherlands Media Art Institute, internationally renowned as a resource for video art. So for Elephants Dream, we’ve had the luxury to challenge ourselves to create real independent artistic content as well. The artists had a lot of freedom from the start; they were responsible for the concept, story and creative development of the entire movie. This has resulted in a lot of quite personal choices, based on what the artists liked to do themselves. I really cherish such an approach, it has resulted in a very motivated team working crazy hours the last months to get it all realized.

But, most importantly; the main target of our project was not only to create a 3D movie short, but to experiment with ways to improve the efficiency and quality of open source development. On this aspect only, this project was just a huge success, and the main reason for our sponsors (the DVD pre-sale) to support it. I know they might have liked a cartoonish funny movie with furry animals better, but for that you get already pretty well served by the bigger 3D animation studios. 🙂

I’m the first to admit that – looking back especially – certain aspects worked out quite weakly; there’s loose ends and questionable decisions, especially in story development and continuity. That’s just the risk of doing experiments, and nothing I regret really. The five artists from our user community who were invited to make the movie were young people with no professional background in filmmaking. Their personal incentive to participate in this project was also to learn from it, and to create a good portfolio for their future career. I’ve witnessed them grow in competence in the past year enormously, something I’m incredibly proud of.

On the technical level, the only major criticism I’ve seen of “Elephants Dream” is the character animation, especially in the opening scene — many reviewers felt that the movements seemed a bit unnatural. Do you agree with these criticisms? If so, what do you think can be done to improve on that level?

Yeah, the challenge the artists set themselves – to use quite realistic personages – is also something that easily works against you. In many animation movies they introduce characters in the beginning in a way you get used to their specific characteristic movements, so you accept a certain level of non-realism easily. (Check the weird walk cycles in The Incredibles for example). Another aspect is that we’ve started work on the first scenes, and ended with the last scenes. I can clearly see the animation quality increase, and that whilst the ending scenes were done in much less time due to time constraints.

We also didn’t schedule to do 9.5 minutes of animation either…. Originally it was more like 6. But, it’s always easier to look back to define the right decisions, eh? 🙂

I’m very happy with the reviews we got so far; luckily the movie was perceived as a professional quality product, and reviewed based on comparisons with what the big studios come up with. Even when we couldn’t satisfy all these quality demands, it has luckily not been branded as a pathetic presumptuous attempt by amateurs!

Do you think there is hope for a full-length open movie project in the near future? Would the Blender Foundation be interested in such a project, or do you intend to continue focusing mainly on shorts?

I’d like to wait a little while with defining what a next project would look like. Given the constraints of “organizing projects to improve open source development”, we might have not much choice either. It would probably mean to work with a new team each time, so most likely be based on shorts only. On the other hand, there’s also clear signals that this approach works well, and creates excitement and involvement of a lot of people, also from producers and sponsors. That might enable us to set up a next project based on larger targets. For a full-length feature film however, we should involve a sufficient amount of experienced film makers as well, and/or invite the first team to participate again. That would put a lot of pressure on the required budget…. You can’t do that based on a 1000 DVD pre-sale target. Would more be like 20,000 or so…. 🙂

How did the process of making the movie feed back into the development of Blender? Are there major technical changes that were made only or primarily because of the film?

Already during the pre-production phase the artists have defined the key targets for Blender development. This then was coordinated with the online development community too. I’ve done the most crucial (re-)development mostly myself, though. Especially on the character animation tools, on the rendering pipeline and compositing tools.

It is especially the latter I’m most satisfied with. In 3D movie production the compositing stage creates a giant content bottleneck. By transparently integrating this in our render-pipeline, a very efficient workflow has been achieved. And, not to forget, Blender now also offers the first production-level open source compositor on the market!

The current summary you can find in our work-in-progress release notes.

What are the key technical features in Blender you want to add or improve for future movie projects?

Depends on what the movie is about! There’s always hundreds of features you can work on. However, we’ll have to work on that anyway, movie project or not. There’s a lot of professionals using Blender now, and they can’t wait for the Blender Foundation to do movies! Look at this studio for example:http://www.plumiferos.com/

I read that at least one proprietary software package, Reaktor, was used for the sound effects. Is this because no equivalent free software solution exists yet? Will future projects have a “free software only” policy?

We’ve limited the “Open Source tools” requirement to our own Studio Orange only. That was what we could keep in control at least, and I can tell you it was not always easy even… 🙂

For sound and music we’ve decided from the beginning to seek an external sponsor. We have chosen to work with the best quality studio and composer we could find, preferably using open source, but not as a prerequisite.

My own competence is solely within the CG [computer graphics, Ed.] side of movie making. When it comes to music editing, or video encoding and DVD authoring, I could only decide to choose to work with external parties with proven competences in that area. I have to be practical in projects like this, especially to ensure it will be realized.

Hopefully, now we’ve got so much attention world wide, we can involve more non-CG open source next time, too. I will definitely strive for the maximum here, but it will fully depend on the amount of professional support we can get.

Blender itself was originally closed source freeware, until it was “liberated” through a fundraising campaign. If you could choose one proprietary application to “set free” where such a goal could be realistically achieved, which one would it be?

Well, the “realistically achieved” demand makes it quite difficult. 🙂 Looking back at similar cases, like Mozilla and OpenOffice.org, it was always very circumstantial. It just happens sometimes, you can’t organize something like this to happen in advance. The only common denominator is “a company in troubles”… so, who’s in trouble now?

What is your personal favorite computer-animated full-length film?

Uuuh… that differs every week! Probably Ice Age (the first one). Mostly because they didn’t overdo showcasing 3D technology so much, but created truly adorable characters and great funny gags.