Author Topic: Motion capture.  (Read 13974 times)

Offline Cor

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
on: January 09, 2008, 06:17:12 PM
does anyone think this is posseble.
i do because it should be the same princeble as te witeboard. i think you nee 1 wiimote for every side. and than infrared lichts on your body. you could have 8 tracking points (ore more if you have more motes.

But i think there are 2 problems 1 how do you kalebrate is. 2d kalabration is easyer than 3D
And does the wiimotes know witch dot thy have to track.

But if you want and can't pull this off. it would be realy cool for indie game devl
tink about it a motioncapture devise ting for only like $120.

Cor.



Offline Shooty McBang

  • *
  • Posts: 10
  • Karma: +0/-0
    • View Profile
Reply #1 on: January 09, 2008, 11:20:07 PM
I don't think full on motion capture would really be practical in a home gaming environment. It could be put to use for other applications, but if you are referring to this as an idea for use at home, I just don't see any practicality. I live in a barracks room and would not have the necessary space to move around as needed. How would you apply full motion capture beyond home gaming?



Offline fALk

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Reply #2 on: January 10, 2008, 02:52:10 AM
The problem with that approach is that tracking point get covered up by other parts of your body - its the biggest problem in this kind of motion tracking - look at the very expensive VICOM system who has after 15 years of research still not figured out how to solve this (the solve it by putting an overpaid person behind a computer for a week to tell the computer which tracking marker is which.)

now using the actual wiimotes to track the movement (like strap the freaking things on your limbs) and we are talking. but you need LOTS of wiimotes to do that - lets see:

3 for each limb = 12
1 for head ) = 13
2 for shoulders = 15
two for hip = 17
and probably at least two for upper body = 19

I do think that for this its cheaper to get a "mocap" suite with the sensors going.



BUT your approach could work for facial tracking where you put about 16 markers on an actors face and have them act out - then this data is used for morphin a 3D model of the face. For that its actually VERY interesting as the whole tracking process would be "perfect" - a common problem with visual facial tracks.
You "only" need four motes for that - that sounds feasable.

By the way the whole tracking market for 3d animations is - while extremely small - extremely well paid for - any solution that would revolutionize any kind of 3d tracking would take off like a rocket - the world is waiting for exactly that. Make a solution even "free" and "open" and you have about a billion indy 3d designers as friend ;)



Offline gabort

  • *
  • Posts: 39
  • Karma: +3/-0
    • View Profile
Reply #3 on: January 10, 2008, 08:34:10 AM
You could probalby set up some limited mocap, but it will be hard. Attaching the lights itself is a pain, and differs from standard techniqe where only reflectors are used to bounce IR light from a strong source to the detectors.

You need to track these points in 3D, meaning at least 2, but preferably 3 detectors. Then you need to process the data, attach the raw data to specific data points, assign the data points to skeletons and then there is still the problem of cleaning up the data, as there are bound to be some missreadings, coverups etc.

Mocap studios are not cheap, but neither is their equipment. I don't think there is a free lunch in this case.



Offline Vicious

  • *
  • Posts: 1
  • Karma: +0/-0
    • View Profile
Reply #4 on: January 22, 2008, 03:50:26 PM
i think it is very possible though limited, and like with all motion capture, its the software and the hardware, i mean Viacom a motion analysis (the two nost bought systems) have very similar cameras but they really differ when it comes to the software, the studio i visited the other day was just using slightly tweaked security camera's. so thinking about how the wii works, if someone made a program to track, and clean the data, and using the wii remotes as camera's, but the limitation would be bad, i think you can use a max of 4 remotes, which the minimal camera i would even want to try a capture with is 8 (and ewww) but 4 is what I'm thinking is the max wii can use, unless i'm mistaken on that but even so, there range that I'm seeing it like 10 ft max, with the 45 degree angle for the sensor, the biggest volume you could make is something like 5 foot by 5 foot, if that, and i have no clue what type of height you would have. so making the right software would be key, but i think i could work, it would be very...VERY hard to work with, but would be neat, especially for someone like me who really want a personal motion capture studio, but doesn't have a 100 g's to spend for the equipment



Offline MikkelJans

  • *
  • Posts: 2
  • Karma: +1/-0
    • View Profile
Reply #5 on: January 29, 2008, 03:33:20 AM
I'm very interrested in creating low-cost motion-capture too.
It should be possible.. The biggest problem is that the wii can only sense 4 dots. But that could be solved by having the light-sources blink fast, one at a time, so the wii's only has to record 1 light-source at a time.
It would only require about 2-4 wii-controllers, maybe the double if you wan't to capture a larger area.

Right now, i'm trying to connect my wii to Maya (A high-end 3D-Software), here's the result so far:
http://www.youtube.com/watch?v=ietyAUD_JQg

I'm using Python with pyBluez on Suse 10.3

I only have 1 wii, so it's still not 3d.. But it'll just be a matter of comparing 2 or 3 wii's data into a 3d-point.


http://www.km-animation.dk
http://www.mikkeljans.dk



Offline Skankintoopiv

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Reply #6 on: January 29, 2008, 06:08:13 PM
well, I have no idea what I'm talking about, but with that whole blinking light idea. If you had them all blinking in a pattern (you'd have atleast 4 patterns, one for each side.) you could set it up to read the data in that pattern for certain areas, and if the dot is not recorded for the spot in the pattern it would have to leave it be. then just have it all set up to a skeleton, so that if there are some misreadings it will not make your arm magically shrink or grow. as in keep the body dimensions the same at all times.

I'd say the minimum would be 4 patterns, and 4 wiimotes if you want it in a 3d space tracking all sides.
I'd suggest about 8 wiimotes, since someone said the range is about 5 square feet. one lower and one higher on each side. still the 4 patterns, so that if the wiimote ontop senses one of the bottom ones it won't throw off the data.

that sounds like a lot of programing, but **LOVE**, what isn't?

P.S. sorry if that was all retarded.



Offline MikkelJans

  • *
  • Posts: 2
  • Karma: +1/-0
    • View Profile
Reply #7 on: January 31, 2008, 04:22:45 AM
Yes, 4 patterns would definitely be more precise, but it's also alot harder to program.
But it might be necessary.. In the tests I've done so far, the wiimote lacks behind when I record at a speed more then 24 frames per second in Maya.

I'll just need to do some more tests, when i get the time to it.



Offline balam

  • *
  • Posts: 58
  • Karma: +1/-1
    • View Profile
Reply #8 on: January 31, 2008, 05:16:37 PM
it an staring idea.
perhaps using the vvvv and wiimote together



Offline fridayblack

  • *
  • Posts: 1
  • Karma: +0/-0
    • View Profile
Reply #9 on: February 27, 2008, 01:04:49 PM
Hi,
I recently created a Wiimote Server for Maya by combining gl.tter's Wiimote Library with Maya's mocap server.   This allows you to mess around in Maya by connecting different Wiimote controls to attributes in Maya. I also created a mel script that sets things up to control a camera with a Wiimote and record these movements. 

here's a short video:
http://www.youtube.com/watch?v=Btrx1PjKeTY

and the website where you can download it:
http://www.parismav.com/tech/wiicam



Offline sjeh

  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Reply #10 on: March 12, 2008, 06:35:26 PM
You may be interested in a short video we've put together demonstrating full 3D motion capture using two Wii controllers.

http://www.youtube.com/watch?v=CT6aQN-lwmo

The accuracy is surprisingly good for such an inexpensive system!



Offline leepmeister

  • *
  • Posts: 4
  • Karma: +0/-0
    • View Profile
Reply #11 on: April 05, 2008, 12:33:17 PM
The accuracy is surprisingly good for such an inexpensive system!

Wow, this is exactly what I need for making a ping-pong robot. Please tell me your code is open source somewhere? :)



Offline johncoths

  • *
  • Posts: 1
  • Karma: +0/-0
    • View Profile
Reply #12 on: October 28, 2009, 03:08:35 AM
I guess there are some sites that provides a technical definition of what motion capture is, how it's used, and why. and more instructions for the motion capture security system.



_________________
Security Systems