Problem-solving

There are always problems that arise when it comes to doing any film project, let alone one as effects-heavy as Morphology will be. Especially things like combining live-action with CGI in the same character, as we’ll be doing with Kadij. You’ve already seen her final draft, but there’s more to Kadij than just the helmet prop. I’m not even talking about the glowy Tron-bits on her. She’s going to have an articulated set of wings, based loosely on the set that Sam used to base-jump to freedom with and that these guys glided around on:

Tron black guard with wings
One of Clu’s black guards with wings

So, the first problem is exactly what these are going to look like. Problem number one is that I have no idea how the wings were actually attached to the Black Guards in Legacy. Problem number two is that Kadij’s wings need to look less sophisticated than the Black Guards’ do without looking like Lego stapled to her back. Problem three is figuring out how they’re supposed to move. And that’s just on the design front.

Part of my process is multitasking. Whether or not it It’s a sort of creative visualization exercise. I’ll have two notebooks in front of me while in class, or have a sketchbook or my computer while watching TV or a movie. I’ll work the left-brain right-brain split, use the right-brain output to free up left-brain processing power. It may sound unintuitive, but it does help. It also gives me a scratchpad in class so that I don’t clutter up my notes with off-the-cuff notes or ideas.

So, the other day, I got this sketched up for Kadij’s wings:

 

Kadij's wings

Sketch of the design of Kadij's wings, including how they form. Click to embiggen

Ironic, that one of the scratch-notes is for a book called “The Myth of Multi-tasking.”

So, a couple of solutions. To the problem of where the wings attach, I’ve decided to grow them out of the identity disc itself, so that it’s more of an extension of the program than some sort of attachment. To make them look prototypical, which is the buzzword with everything in this flick, they’re more angular, less refined, less structured than the original and will open more like a switchblade than a motorized system. And they’re less efficient, bleeding off some of their energy into a sort of heat exhaust. I’ll post WIPs of the 3D model when I have them.

All of this talk about prototypicality is really getting my brain-juices flowing about the Recognizer. I think that I’ll tackle that again soon. It’s really been kicking my ass.

Another problem that I’ve had to solve is tracking Lesley’s head and body for these CG costume elements and props. The problem stems from simple limitations in the software I have. I don’t have a 3D motion-capture solution, just After Effects’ built-in motion tracker tools. They only work in 2 dimensions, so I have to figure out a way to use AE’s 2D tracking and a bit of 3D spoofing to get the elements to move properly in 3D space, which is essential because they’re so reflective. They have to properly raytrace everything in the scene, otherwise the illusion’s thrown. So, while waiting for a friend after class the other day, I drew up this:

 

Mocap solution

Sketch and early maths for motion tracking solutions. Click to embiggen

I realized as soon as I put the sketchbook away that I’d made a mistake. I can track Lesley’s face, for sure. I can use that to fix the wings to her back. The same method can work for both CG costume elements. I can get the tracking data into 3D Studio Max, and then link the helmet to that track data.

But.

As soon as Lesley moves towards or away from the camera, the tracking data, because it’s done in two dimensions and not three, is going to be wrong as soon as I start moving the helmet model in 3D. I’m not yet sure how to correct for that. Minimizing her coming towards and away from the camera is one option. That’s simple enough to do, but feels like a cop-out. Doing a manual track is another option. That would be more time-consuming, but would work. I guess we’ll find out in the post process, but if I can figure out a way to compensate for it mathematically using some sort of script, I might do that. If I decide on an origin point, and record how far she travels in what amount of time, I might be able to scale the tracking data itself as a function of time.

I’m thinking on the page now. Probably not scintillating reading, but such is the nature of the spillway. I spill. If you have any insights, I’d love to hear them!

Paddle your own canoe, guys and gals.

-Trevor

One Comment

Leave a Reply