Interview with Robert Hödicke
Thank you for taking the time. What the Wonderlamp Djinni is, I could already read about it on the web. But please tell us in your own words what it was about.
Sure. The project was very ambitious. It was about developing software to produce animated films faster, but also in a very different way - more like how live action films are produced. That users would be able to program the behavior of their characters, I'll say, with a relatively small amount of effort.
When you mention something like that today, people immediately think of generative AI. You can make movies with it, after all. Did you guys want something comparable?
On a very abstract level, you could say we had the same goal. But we took a very different technical approach. For us, it was about developing a tool that actually left all the creative decisions up to the director. Our ambition was to broaden the players' range of decision-making, because conventional animated film production is very rigid due to its specific workflow.
And one ingredient of that system was Inverse Kinematics: If the foot is in contact with the ground, it mustn't slide, and then I have to move the whole skeleton accordingly. Or: The character steps through this door and then he goes to the window on the other side of the room. Then you can check with Collision Detection, he can pass to the left of the table or to the right of the table. The system should then suggest a variant, and the director can say nope, I don't want him to pass to the left, I want him to pass to the right.
This is a mechanical problem where a background in physics comes in very handy. And we didn't have that. Yes, so of course there is literature on how to do that. But if you want to implement such a thing quickly ...
... then you need a physicist.
Yes, and not only a physicist, but the second problem was: We had implemented the system in C++. You can do fantastic things with C++. But it is a language that is very complex, where you need several years to really master it. And Ambrosys had someone on board who was a physicist by training, but who was also an extremely high-profile C++ developer who co-hosted the C++ conference here. A proven specialist. And that's why it went so swiftly. We didn't need to spend a lot of time talking about the details; I think he immersed himself in the issue in two days and then made concrete announcements about the effort that would be required, and then simply delivered it within three weeks.
Now, in the style of that infamous job interview question: Say something bad about Ambrosys.
I can't think of anything, unfortunately.
Don't you have a few nice anecdotes?
Yes, but they're not really anecdotes about Ambrosys. The biggest problem was that I underestimated how hard it was to find qualified personnel. As CTO, on the one hand I was the architect of this system, and on the other hand I was representing the team to the outside world, to the rest of the company or the customer or the owners and investors. So I felt reasonably stretched between those roles. Somewhere I heard that once "I could be a so much better manager if I had a better team". Of course, that's a cheap thing to say ...
You need a team like Brazil when they won the World Cup in '94 with a team that they say could have done it even with Erich Ribbeck.
Yes. Sort of. And the developer of Ambrosys was, so to speak, the employee I would have loved to have, where these problems didn't arise and where you could simply agree on what had to be done and by when, and then it just happened.
Now you have already been around quite a bit in the business. To whom would you recommend Ambrosys?
Ambrosys has very specific, very very relevant qualifications. Markus is, after all, a professor of statistical physics, and there are many things unfolding right now that can really only be grasped in statistical terms. There are so many more things that have a physics component to them, where I would think it would come easy for Ambrosys to develop software.
Now you're on to physics again.
OK: Did I mention that Markus plays the saxophone better than I do? No, seriously, let's highlight one other thing: It's this autonomous, self-reliant work at Ambrosys. So that you don't make just some piece of software but understand what this component needs to do in context. And based on this understanding, the countless decisions you make every day as a software developer actually make sense.
So that you, as a customer, don't have to micro-manage.
Exactly. I certainly have my basic skepticism about subcontractors. But at Ambrosys, the aspiration was there to deliver something that actually satisfies the customer and not just something that formally meets the spec sheet.
Thank you, Robert. It's not every day that you get to meet such a versatile and interesting person. Good luck with your current projects.
The same to you! Have a good trip back to Potsdam.