Hey all,
This is a really, really cool idea - I'm glad that someone has finally taken the desktop mocap idea to task. Our company is currently in the development process of a 6DOF full body mocap suit which uses homebrew IMU (inertial measurement units) for joint tracking, similar to the Gypsy Gyro suit. We're looking at a full body price tag of ~$10,000, compared to Gypsy Gyro price tag of $80,000. This system is nice (as opposed to camera tracking) as there is no feature tracking necessary, which is the major difficulty with full-body visual mocap.
The OptiTrack product looks like it might be another option for us, and other small game companies. We'd like to be involved in the beta process as well. We have an extensive background in C++ Maya plugin development and look forward to trying out the OptiTrack API.
If for nothing else but facial feature tracking, we'll use the system.
The major issue we see with the OptiTrack system is data merging from multiple cameras. It's unclear to us if multiple cameras' tracked points or "features" can be merged into a 3d cloud of tracked features. This is hindered by feature occlusion (not all markers visible at all times), and depends heavily on the number of cameras used, and the placement of the cameras.
Does the API provide the facility to merge "N" number of cameras' 2d data sets into one cloud of tracked 3d features? I assume if not, this is a work in process for the API... This would be necessary to output data which could be imported into Motionbuilder or similar mocap application.
Finally, a tool for determining placement of individual cameras encircling the stage would be necessary for good data output.
At any rate, great work so far - we know you've probably discussed and/or addressed many of these issues already. We're definitely interested in becoming a part of the development/beta process. Any ideas on product timeline for the beta?
Cheers,
Erik
EGirl Interactive LLC
[ July 02, 2005, 02:13 PM: Message edited by: egirl ]
game mocap and software questions
Re: game mocap and software questions
Hello:
Thanks for the feedback, glad to know you might find the system useful.
The current OptiTrack system is really designed to be a Development Platform, where users can create their own optical tracking systems, for whatever application they might come up with. Motion capture is one of the key applicaitns we envision for the system. Currently you can do things like facial motion capture, with a single camera, very easily, and it works very well, resolution better than 1mm.
You are correct, doing the mulit-dot and multi-camera dot corrolation is a hard task. We are working on this and do plan to have a "tool kit" available in the near future, some months away. This will allow for multiple cameras in a user defined array with point tracking and occlusion management. It will be quite a task, but I think we are up to the challenge. Meanwhile, we encourage users to play around with the cameras, as they are cheap, and see what cool things you can do with them. The SDK is good and easy to use, so making your own apps is not terribly hard.
Thanks for the feedback, glad to know you might find the system useful.
The current OptiTrack system is really designed to be a Development Platform, where users can create their own optical tracking systems, for whatever application they might come up with. Motion capture is one of the key applicaitns we envision for the system. Currently you can do things like facial motion capture, with a single camera, very easily, and it works very well, resolution better than 1mm.
You are correct, doing the mulit-dot and multi-camera dot corrolation is a hard task. We are working on this and do plan to have a "tool kit" available in the near future, some months away. This will allow for multiple cameras in a user defined array with point tracking and occlusion management. It will be quite a task, but I think we are up to the challenge. Meanwhile, we encourage users to play around with the cameras, as they are cheap, and see what cool things you can do with them. The SDK is good and easy to use, so making your own apps is not terribly hard.