Interview with Ton in Linux User & Developer Magazine

Last month we mentioned that issue 68 of featured an interview with Blender’s lead developer Ton Roosendaal. Editor Daniel James then promised us we could publish the interview text later on and he kept his word: here’s the full interview! Thanks, Daniel!

3D Dreams

Ton Roosendaal is the creator of Blender, the leading free software and cross-platform 3D tool used by countless artists and animators around the world. As the Blender Foundation begins to plan its second open movie project, Roosendaal tells the story of how Blender freed the source

My background was in industrial design, and I worked as an independent designer through the 80’s. I also had a technical background, so I liked to play with computers. In 1985, I bought my first Amiga, which had an amazing system with colour, and painting software - and you could do things with video. The Amiga also had a couple of 3D tools, and that’s how I got in to 3D.

A couple of years later, I decided to start an animation company, doing video graphics, video animation and effects, mostly focused on 3D animation. It was called NeoGeo, or ‘new shape’ in Greek. The Neo-Geo was also the name of a games console, but that was a coincidence, and happened later too. We established our company in 1988, and a couple of years later the games console hit the market. It was difficult for us; people got confused with the name, especially if we did things for 3D games.

Later on, that work became the source of Blender, because in our studio we did a lot of our own software development; mostly because we didn’t have money to buy all the expensive software. Also, we had switched to Silicon Graphics systems, because by the early 90’s the Amiga was really dead. The Amiga still had an enormous advantage over Windows systems at that time, but Silicon Graphics systems were what the 3D industry was working with. So we decided to develop our software on SGI, completely ourselves.

Blender was actually created in 1995 - it was the third generation of our own tools. Completely written from scratch, it was designed to be the in-house production tool. By then we had six years of experience in computer graphics, and so the design for Blender was based on everything our own artists needed. Most of the development was even done by the artists, and that’s what makes Blender unique. It’s not a marketed tool; it’s not based on easy learning. In a company, you can discuss with the artists how things work, and what those people want is extremely fast workflow. Once you get into Blender, it really flies. You have to grow into it, or the software has to grow on to you.

At the end of the 90’s I decided to quit the animation studio, mostly because there was no real development in it. The Dutch media industry is pretty much corporate; we don’t have a big movie industry. If you want to do animation work for movies, you have to go to England or you have to go to Hollywood, and that was not an option for us. So I decided to continue with Blender itself. The tool was great, and I thought it could be an interesting thing to work out how to market it. In 1998 we published the first public versions, including the first Linux port, and opened a website. It was freeware, back then; you could download it for free, but it was not open source. Because it was the first relatively-professional 3D tool for Linux, it became incredibly popular.

A dot-com phoenix

If you did something with the internet community and free software in that period, it also attracted attention from investment companies and venture capitalists; it was the dot-com era. In the Netherlands, we had a big bang when one internet provider went public. It was valued at something like two billion euro, and only had 100,000 subscribers. Every subscriber was worth 20,000 euro - that was ridiculous!

So I thought ‘OK, let’s give it a try, you never know!’ We got funding and set up the company Not a Number. We wanted to keep the core of the software freely available, and develop professional services around it. But the month we started, the internet bubble burst. After that, the people who had invested in Not a Number got a little bit scared, and all the beautiful plans had to be condensed into a very short period, because now we had to make money - and that was quite difficult. In 2002, Not a Number closed for good. We were making some money, but it was not enough to be self-supporting.
Then we had a couple of months that were a really big black hole. I managed to get the investors in the company, who by then owned 90% of the shares, to agree to make Blender open source, licensing it as GNU GPL. They agreed on this under two conditions.

One was the licence, because I told them that the GNU GPL was the most radical and strict licence in the market, and was ‘barely commercial’. Of course people can do business with it, but the investors were afraid that other companies could take the asset and create commercial software out of Blender. The GPL is very strict about mixing proprietary software with software that is under the GPL. So the chance that any company would pick up Blender and market it was pretty minimal.

The other condition was a fee of 100,000 euro, which wasn’t a lot compared to what they had invested, but it was (most likely) an amount they wouldn’t have received if they had sold Blender to another company. Surprisingly enough, we got the money in only seven weeks - it was an amazing time. For the Blender community we had a sort of membership, a ‘founding member’ ticket for 50 euro. You could download some extras and goodies, and that membership had some kind of status. That’s what we sold most of. We also had a lot of publicity, on Slashdot for instance. Every Slashdot posting brought a lot of donations, but these were all small - five, ten or fifteen dollars. I think in total we had about four or five thousand dollars in small donations, and a couple of big ones - we had $2000 from one person. Most of it was from the community; there were no corporations involved.

The Foundation

The Blender Foundation was established, to sign the deal with the investors for the licensing. When we published the sources we had to set up services; CVS, mailing lists, making sure that development can happen. That was during the first year; it was very interesting to witness how this process started. Nobody knew the sources, and they were hardly documented. They were completely stuffed with Dutch comments in the code. Why would you do comments in English if it was never the purpose to publish the code? You need a lot of work to get that code into a shape that fits better for open source development.

I also decided to be at a distance for the first six months, to give people the opportunity to mess around, otherwise everything would have fallen back to me. It’s important that people should feel invited to participate. Even if they do really bad, horrible things, you have to allow them, because that’s the way they learn how the software works.

It was also great to watch the dynamics of an open source project, really embracing the features that people would like to see in the software. It’s quite different from a corporate approach. A lot of work has been done on the interface, trying not to lose the workflow for power users, but trying to get more access for new users. A lot of people thought that was important, so it was just added.

The other thing that was surprising was that the gaming part of Blender didn’t get a lot of attention. Most of the focus was on the core tools, like modelling and rendering. All our ideas about having an interactive creation tool - Blender still has a game engine built in, and game logic - were only really interesting to a very small minority of users. It’s still so; there’s an active group working on the game engine, but it’s small.

The core Blender development team, those that have write access to CVS, is now a little over fifty people. Around that is about one hundred people who provide patches. Of course, because Blender has a very large user base, we have thousands of people providing test reports, and testing builds. That whole cycle, of making stable software releases, is something we do really well. Blender has been getting a good name as a stable platform to work with.

It’s impossible to estimate the size of the user community; we don’t have user registration. Regarding our own distribution from the blender.org website, if we make a new release about 200,000 people download it within a month. That’s the core of the user base, who want to have the latest version. Then there are mirrors, and most (if not all) Linux distributions have Blender. All of that we don’t count. Then there is download.com, and Macintosh websites; they all distribute Blender.

Dreaming in 3D

Because of my background in an animation studio, I knew that when developing a 3D tool with the complexity of Blender, it’s important to be very close to the artists. They are the people who should define how the software should work. That’s very difficult to extract from the thousands of feature requests on a website. What would be the most wanted or most useful addition to the software? There are a lot of things you could work on. From my practise, I know that if you give a team an assignment, of making a movie short in six months, they will quickly find out that maybe 90% of the feature requests are nice, but you don’t need them to make a movie. So the artists quickly go to the core, key things that you really have to fix, otherwise you can’t make the movie. That focus is incredible, and also inspires the developers to help this small team of artists to get it done. That’s why I would love to have more projects like Elephants Dream.

We started the Elephants Dream project by partnering with a Dutch art institute that specialised in video art - very crazy, weird stuff, like dripping water for five hours. They also challenged us to make something that had an abstract, artistic level; experiment with alternative or non-linear storytelling, develop our own concepts. That’s what I’m really proud of, because I worked first behind the scenes to get the whole organisation set up, including the financing. So before I even announced it in public, I knew that this project was going to be realised. Because of the financing, there was no burden for the team to make something commercial, or something that would sell. It was up to them - they could decide what to make, and it was their personal vision.

I talked at SIGGRAPH this summer to a couple of people from the movie industry. The biggest compliment they made is that they said ‘Wow, it looks good!’ The animation quality, rendering quality, modelling, looks like it’s been done with a professional product. For Blender, that’s a big leap forward, definitely. A couple of people criticised the story, the editing; it’s sometimes a little bit jerky. But that’s OK - they should have been there during production! The team put six people together who didn’t know each other before. They had to finish something in six months.

Before you have a team of people who can work together, it takes a lot of time. Actually, the movie had to be made in two, or two and a half months. The first three or four months we used for design and ideas, getting up to speed and getting the software ready. In the end, people were making one or two minutes of animation in a week, which is a lot. I would have loved to keep the team together for another year; then we could have made a whole movie!

Also, it was obvious at the end that the scenes we did at the beginning lacked quality. It’s a luxury we didn’t have, but you should go back to the editing room, cut the movie in pieces, establish it again, and make some more scenes. But we only had a limited amount of money, so that was it.

Preparation time isn’t usually included in the production time of a movie. Most movies start one or two years before the production starts, on the script, storyboarding and pre-visualisation, that kind of thing - you can do everything right. We had to do all of that in six months. For the next project I will try to have better pre-production, so that when the team is together we can get into actual production. It depends on a new partner, how long we can work on it and how big the team is.

Remixing the source

Not only the software we used was open source, but the entire studio database had been published (under a Creative Commons licence). That’s crazy, and has never been done before. I think we deserve the claim that Elephants Dream is an ‘open movie’. We did have some remarks about the audio part, but then Blender is a computer graphics project. We couldn’t afford to include open source audio, and I have no idea how to manage it. We found a studio sponsor for music and the soundtrack, and it did use open source in the studio, but not exclusively, because there was no open source audio development involved. We might do that for next time, but it would have to be organised by an open source audio project, that would participate and help out the team to get it done.

Also, because we wanted cinematic quality, everything was rendered in 2K high definition; the audio soundtrack had to be 5.1 surround, Dolby, theatre quality. That’s still little bit of a problem with open source production. I hope sometime it will be there, and we can support that.

Another nice thing with making this movie open and freely available is that it has become a sort of reference standard for high definition television. There was a big broadcasting convention in Amsterdam, with five or six big halls, the whole European broadcasting community presenting for ten days. I heard that at least ten different booths were showing Elephants Dream on high definition TV. That’s mostly because there’s a big lack of high definition content. You can do it with a camera, but cameras are noisy and it doesn’t look that crispy. If you have the original 3D renderings we published from Elephants Dream and you show them on a high definition screen, it looks so crispy.

The first European HD-DVD title was Elephants Dream; a German company published it. It’s a commercial product but we don’t get anything from it - that’s fine, it gives Elephants Dream a bigger audience and distribution. I’ve lost count of the number of TV stations that have shown Elephants Dream; I just heard that the main Argentinian TV channel showed it. It’s impossible to keep track of everything; the same for festivals and awards.

The moment we finished the project we had to move on to other things, and since Elephants Dream is open and free, everyone said we should submit it this or that festival. I said ‘Please submit it yourself! I can’t do all of that, it’s too much work!’ So a lot of people have been submitting it themselves, and we’ve had a couple of awards.

What’s next for Blender

At the moment, we have to focus on the next release of Blender, and there a lot of loose ends from the Elephants Dream project. There are things that were added in the Google Summer of Code project that we have to get into a stable release. I would like to migrate a few more of my responsibilities, as there’s still too much depending on my availability. I’m also the person who likes to organise things and get things done, so I can’t do the next movie project if I’m developing too much.

One of the things we’re going to do after the next release is a complete re-code of the internal event tool system, customisable hot keys, and Python integration with the event system. Then I hope in the Spring, we’ll be ready for the next Open Movie project, which like the first one will be a temporary project in a studio for half a year, or there’s a possibility it will become permanent. An open studio where artists and developers can come over to work on independent cinema, using open and free software. The Mozilla style of corporation could work well for computer graphics, especially because there is a big community of developers, but most importantly because developers could come over for just two weeks or a couple of months to work together on specific areas. They won’t be able to give up their jobs or their lives, but they might be available for a short period. Even though the internet is great, and software developers love to work via the internet because they don’t get disturbed, meeting in person is always much better.

That kind of central hub where people can meet and work on projects together is something I think we really miss. Maybe we can even expand that to other open source projects like games, or video editing. There are a lot of computer graphics open source projects that find it difficult to get into the commercial or professional market. I don’t know why - most of these projects are being organised by technical guys which are happy when the software works, when it compiles, or when they can do cool things. They’re not so interested in going to trade shows or to give talks, to make sure they get reviews in magazines and that kind of stuff. The Blender Foundation is doing that pretty well, because of my commercial background. I don’t mind organising things, setting targets, or having deadlines.

[@more@]

--------------------next---------------------