Tuesday, March 27, 2012

Vicon, Xsens, and Kinect

We head-to-head tested the capabilities of the Vicon, Xsens, and Kinect mocap systems to assess the pros and cons of each piece of hardware. Markers cannot stick to the Xsens system, hence the incredible tape job. In conclusion, Vicon had the smoothest, most accurate capture, but it's also by far the most expensive, and marker setup/occlusion can be a headache.

Xsens did a great job tracking rotations. Not having a stage or a long setup is a plus. That being said, it had a bit of trouble tracking translations (if I were to run around the room and return to the middle, the character's position could be slightly off). This is a perfect rig for game animation cycles. It is also significantly cheaper than Vicon, but still quite expensive.

The Kinect System worked surprisingly well, so long as I continued to face forward (We only had a single camera setup for it). It was a little bit jittery, but for the price of ~$200 it totally beats out the competition in affordability.

Tuesday, March 13, 2012

Hand Joint Rigging MEL Script

Yesterday I was getting bored of rigging all 15 finger bones in each hand to curl, so I wrote a script that will expedite the process.
Basically, it relies on very specific joint naming conventions to construct driven keys that operate each finger numerically (the channel editor on the right). The whole point is to get the fingers to open and close with precision and ease, so I'm happy how the script turned out.

The naming convention to get this rig to work is as follows: 
  • Name the controller (in this case, just the joint wrist) LeftHand and RightHand, respectively
  • Left Index joints 1, 2, 3, and end: index00_L, index01_L, index02_L, index03_L
  • Left Middle joints 1, 2, 3, and end: middle00_L, etc.
  • Left Ring joints 1, 2, 3, and end: ring00_L, etc.
  • Left Pinky joints 1, 2, 3, and end: pinky00_L, etc.
  • Left Thumb joints 1, 2, 3, and end: thumb00_L, etc.
  • Right fingers are the same except ‘_R’ instead of ‘_L’, obviously.
You can download the script here.

Thursday, March 8, 2012

Photofly to 3D Printing Test

I was thinking about experimenting with the scanning software 123D Catch (in beta) to see if I could print figurines from photography with relative ease. Jared was my first brave volunteer, and he struck this pose for 5 minutes while Madeline and I took photos of him from 360 degrees (about 70 photographs total). We uploaded them to the program and it rendered Jared with acceptable detail. I spent an hour cleaning up the model and preparing it for Zprint.

I will be painting it some time after Spring Break. Then I will make more hilarious models of various people.

Friday, March 2, 2012

491 Mech Brawler Prototype Animations

Although production for 491 will not formally begin until we take the class next semester, we've been determined to push its approval through during the pitch session in April.

Therefore, amidst the chaos of GDC and spring break, we are building a prototype! It requires placeholder assets, typically boxes, so I thought "why not just make the guy out of boxes?"

(Suited performance courtesy of Chris Oslund, our producer.)

It didn't take that long at all (a day to mocap and a day to model/rig the character), in fact the longest time spent was waiting for Motion Builder to boot back up after it crashed for the hundredth time.

Now that I think about it, Motion Builder is kind of a troll program. The "export" command renames the current file opened, so if you unwittingly change stuff and save, you affect a file you thought you sent off down the pipeline.

But to end on a happier note, everything is getting done faster than anticipated, and better than expected! :)