The following article explaining why Remedy recently posted a photo of a dog in a mocap suit to Twitter was submitted to Gamasutra by Remedy Entertainment's Vida Starcevic, who adds that "it was all done in collaboration with our amazing animation and cinematography team. (And Uuno, of course.)"
We had a dog, a mocap studio, and some spare time. So we thought, why not?
Let's try and see if we can capture some useable data with Uuno, the Staffordshire Bull Terrier belonging to our Junior Cinematographer, Sami Kastarinen. That's how it all started, at least.
And then pictures of Uuno with mocap markers stuck to him were published on Remedy's Twitter, and a spur-of-the-moment experiment went viral.
People wanted to know who the dog was. They were speculating that we're working on a John Wick game. (We aren't.) Some of them couldn't believe we didn't stick a marker to the tail. But all of them agreed on one thing: Uuno was a very, very good boy.
Here at Remedy we have our own, custom-built motion capture studio, and we pride ourselves on doing a lot of our own mocap. However, we've only ever captured humans. According to Henri Blåfield, our Animation Team Supervisor, the team did what was perhaps an even less serious test with a canine some years ago. So this time, we decided to try again.
Because of his playful character and fast learning skills, Uuno was the ideal candidate. The only goal we had in mind was to do it just to see if it was something that could be done with the equipment we have. Maybe, we thought, if the data worked we could push it forward, put it in Motionbuilder and see if we could do anything with it.
First came the planning and the logistics. We looked at some reference pictures and data from other studios just to see what their tactic was for doing motion capture with dogs. Our baseline was a human, since that's the kind of motion capture we do at Remedy, and we looked at the movement differences with joints and specific areas of the body between humans (or bipeds, if you like) and quadrupeds. In our case, Uuno the Staffy.
Since this wasn't anything that was preplanned all too much, we didn't have the time or the resources to sew a custom-made mocap suit for Uuno and stick markers to that. We decided to improvise further, and so we borrowed a surgical recovery suit for dogs. Uuno had never worn one before and although he's a very patient dog, the thickness of the fabric made him feel uncomfortable and weird.
Luckily, Uuno had his own walking harness that we could use as a base for the markers. Additionally, our Junior Cinematographer Sami and one of our animators, Matias Leminen, added markers with tape that we knew were going to come off easily. Of course, we wouldn't put tape on a dog if it wouldn't come off easily! The amount of markers we used for this were enough to give us data for a baseline. But ideally for future reference, we would add more. For instance, we could have added a tail marker.
However, this time around we felt that it was more important to get natural movement out of Uuno rather than putting the markers on exactly the right spots. When the first marker set was placed, what we noticed was that Uuno was walking a little stiffly so we needed to adjust our marker placement. To achieve that natural movement, we needed to get Uuno even more comfortable with the placement of the markers.
When we got Uuno comfortable with the markers, we started recording some baseline walking data. We also recorded a bunch of other stuff, like tricks that Uuno can do: rolling over, doing the squirrel pose, that kind of thing. Overall, the situation was new for Uuno and we felt that he was excited and had a lot of fun. We saw that there was potential in him, but for this first test we wanted to keep it light and playful.
As for Uuno? He had a lot of fun doing it. No dogs were harmed during this experiment! He likes attention... and tasty chicken treats. Comparing this to our earlier motion capture test with a quadraped, in this case Uuno was bigger than the dog we used last time, but the issues encountered were quite similar to the ones we had previously. That shows really how much effort and training you should put into a case like this.
Looking forward, if we at some point find ourselves needing motion capture data from Uuno, the first thing that we would require would be a proper motion capture suit made of a lighter fabric and with solid marker locations. While it's our first priority, the suit is also a big, big if – what if Uuno never fully gets used to it and his movements are unnatural? Then the data might not be useable in the long run after all. At most, it would be used just for reference.
What kind of reference? Well, the data we would get out of successfully doing motion capture of an animal could be compared to keyframed animation, so that we could see which of the two solutions would be more cost effective for us and our needs. Are we dealing with a gimmick? Do we need the animation for a playable character? A sidekick? The amount of data we require depends hugely on the kind of character we'll be using it for. Depending on how much we need, motion capture could be the solution, or perhaps keyframe animation would still be better and faster. We can't know unless we try.
Even if we don't manage to record motion capture we are happy with, we could still use the data as reference for poses and timing. There is also the question how many technical loops we would have to jump through to get this to run in our Northlight engine.
If there's a big enough need to record motion capture of Uuno in the future, we think this is viable and could be possibly pushed forward. Remedy just moved to a new, bigger studio in Espoo (check out our work in progress video) because our company got too big for our old one, and a bigger building also means a bigger motion capture studio.
While our new mocap studio floor is about four times the size of the previous one, we were not satisfied with just a larger area. We wanted to go one step further in the quality of the data we capture. This naturally meant that we needed quite a few more cameras to meet that goal. With more cameras, we can make sure that our capture data is more solid. A larger volume size gives us the possibility of capturing longer movement ranges and choreographies.
And who knows, maybe even a dog.