Hello!

I am developing a motion capture application:

http://www.geocities.com/mocap_is_fun/

Currently it can output bvh files to be loaded by Poser:

http://www.geocities.com/mocap_is_fun/Poser.wmv

Well, If you need a beta tester then you know that I am available.

The poser wmv looks great, by the way.

great project!

Please keep going on this and keep us updated. This is exciting!

-jjsemp

Sure will.  By the way what kind of motions do people want to capture?  How do you intend to use the motions for?  Anyone interested in facial capture also?

Quote - By the way what kind of motions do people want to capture?

Dance motions.

http://www.zippyvideos.com/2447054825293786/risky12b/original

This is the most exciting Poser development in years.  Are you suggesting that it is possible to make Facial-motion capture as well as body motion capture?  If you can pull this thing off you'll be able to retire at a very young age.  Good luck and please keep us all posted.

This really is awesome.  Can't wait to see where it goes from here.  I've said before that DAZ should sell a mo-cap system because the Poser/Daz Studio community would really benefit.

Great work!

Can someone make this a sticky?

The smoothness of the cap in the last .mov is impressive! Those optitrack cams are reasonable as well. Any idea on how close you are to having a 1.0 version?

Quote - Sure will.  By the way what kind of motions do people want to capture?  How do you intend to use the motions for?  Anyone interested in facial capture also?

What I'm most interested in is purchasing your software when you're done and setting up an inexpensive system of my own.

The problem with "canned" motion capture moves-for-sale are that it's hard to find exactly what you're looking for when you need it. I have limited use for dance moves or fighting/kung fu moves, which seems to get all the attention from mocap vendors.

What I really need are little ambient things that nobody ever thinks to sell. A few years ago, for instance, I was on the lookout for a simple walk and 180 degree turn around. Couldn't find a good one anywhere.

So I hope you make your system available for sale at a reasonable price. I'll be first in line to purchase it.

And, yes, facial animation would be fun, but please get the full body mocap finished first, since that's where the real need is in the marketplace.

Great work!

-jjsemp

The wmv looks good.

What kind of motions? I'd like to have a library of short BVH with common actions that could be built up into larger animations. Mostly every day things like walking, start walking, stop walking, turn a street corner, move from wlak to run, run to stop, kneel down place object on floor, pick object up, lay down on bed, lay down on floor, sit on chair, walk up stairs, that sort of thing. The most important point is that the motions should integrate, one BVH moving seemlessly into another. I'm talking about the poses intergrating here, not the spatial locations/orientations, that part can be overcome by repositioning the BODY actor.

Followed by the rest of the Poser Animators, no doubt (according to legend, that's 12 sales right there, as there are only supposed to be 12 animators using Poser... ;) ). Those cams aren't bad at all, pricewise. Does your application support only single figure capture, or more than one? And I agree with jjsemp on the full body functionality first. Facial is going to run into the lack of boning and wide variance of morph behavior in the Poser world real quick.

First goal is non-realtime single person full body capture with relatively small capture area. I have been testing with mostly stand up motions, I'll need to investigate for laying down type of motions where markers will be hidden from the views of camera.

For facial capture, I thnk I can just export the 3D marker positions.  Receiving software will have to assign them to mesh vertices.

yosh, I think they may not have noticed the DIY acronym. apparently they're thinking you will be doing all the mo-caps, then distributing bvh files. is it your intention to write software for which the users would require 6 of those cams? I don't have any of those, but I have others, and they can be rather expensive.

Yes.  They(NaturalPoint) have a sales going on, if you buy 3 or more OptiTrack cameras they are $200 each.  So you are right when you get six of those and other equipments like usb hubs/cables, tripods, total cost might be close to $2000.  That is why I am a little hesitant about releasing my software.  I don't want people to spend that kind of money until my software works perfectly. 

By the way I think it is still much cheaper than current commercial systems.  The cheapest optical mocap system I found was metamotion's MotionCaptor which is like $45,000(http://www.metamotion.com/captor/motion-captor-comparison.htm)

$200 is a good price for the OptiTrack cams. as you know, they also sell software to convert the camera data, but I'm guessing you're using their SDK to write your own, so these people wouldn't have to buy their software. anyway, my experience is that 4 metres is a good maximum range for these things, unless one uses auxiliary IR sources.

Well, I saw it.... :P And 2 grand is -not- hideously expensive, considering that Gypsy is around $12,000 for starters...and the Gypsy Gyro is $20,000 last time I checked. Not to mention you're talking about a mocap rig that could go anywhere in a small suitcase and a laptop bag. Admittedly it would take me a few months to get all the gear together, but to be able to create my own mocap files for those shorts I keep starting (crafting cardboard primitive mockups of Davo's SFCS consoles would be easy for example, giving a solid surface return for the actor. Then it would just be a matter of a little fine tuning on the scaling, and you've got motions that don't float over the console). Yoshi, how are you handling the marker occlusion issue?

that's why there are 6 cameras, dale. if the body wearing the markers were transparent, one would only need 3 cams.

Don't forget, some of the 12 of us are Mac guys...

bob, don't forget about bootcamp or parallels workstation (at least not until all the reviews are in)

I'm not on one of the newer Macs so those aren't relevant yet. Sigh...

Quote - how are you handling the marker occlusion issue?

Having more cameras is a good solution as long as the host computer can handle the incoming data.  My PC is Pentium4 3Ghz.  When I have six OptiTrack cameras connected capturing 30 markers it uses about 60 to 70% of CPU. Also having more markers can prevent occlusion.  At least three markers(two if immediate parent bone position/rotation is already known) is needed to define position/rotation

Quote - Don't forget, some of the 12 of us are Mac guys...

Sorry my software won't work on Mac.  This is because there is no OptiTrack driver for Mac and also I am using DirectX for 3D rendering/calculations

This looks great! Though your software won't work on a Mac, is there any chance it could output Mac-compatible files? As a Mac guy, I've been able to translate certain PC files to Mac just by doing a simple search-replace edit in Word to get rid of some paragraph returns, so it seems like it might be possible to have your software output Mac files. Keep up the good work!--T (apparently one of the 12 Poser animators)

Waaaaaaaaahhhh!!

I want it!!

When!?!

How much?!?!

Quote - This looks great! Though your software won't work on a Mac, is there any chance it could output Mac-compatible files?

I haven't used Mac version of Poser.  Does it load bvh files like PC version?  If so it should work with output from my software.

Quote - Waaaaaaaaahhhh!!  I want it!! 

When!?!

How much?!?!

horndog40, are you Chris?  If so I just responded to your email.

The Mac version of Poser does load bvh files, but bvh files seem to come in many flavors. I've had some that work fine, some that won't work at all, and some that I can mess with a little to make them work, so I was curious if your software would make a Mac compatible bvh. Is there any chance you could make a bvh file available to us Mac users to test?

Thanks--T

PS: what was your response to horndog40? I'm very interested in this system--I've been looking for years for affordable mocap, and am in the middle (well, first 20%) of a long animation project I've been working on for 1.5 years so far. Even purchasing a PC just to run this system would still make it far cheaper than any other system out there.

Quote - Sure will.  By the way what kind of motions do people want to capture?  How do you intend to use the motions for?  Anyone interested in facial capture also?

I have a wide range of motions to capture, from simple turning, walking and stopping to fighting scenes. I would use them in animated narrative films. Facial capture would be awesome, but not critical, because Mimic is available for a lot of that.

I'm also interested in beta testing if and when it happens. Please keep us posted.

Also, I don't know about everyone else, but I would probably NOT pay 2 grand for this system, simply because I could take that 2 thousand and book a mocap facility to get everything I'd need for a particular project. Yes, the other systems you mention are 40 and 50 grand, but that's because their clients are high end gaming companies and corporate movie studios, who will pay for those systems with one production. I don't have those clients and would only be using this for my own personal projects, in much the same way I use Poser, where the possibility of recouping any outlay of money is slim. If a piece of software or a utility isn't extremely reasonable (500-1000 max), I won't be buying it, I don't care if it can make my coffee and cook my breakfast.

Quote - The problem with "canned" motion capture moves-for-sale are that it's hard to find exactly what you're looking for when you need it. I have limited use for dance moves or fighting/kung fu moves, which seems to get all the attention from mocap vendors.

What I really need are little ambient things that nobody ever thinks to sell. A few years ago, for instance, I was on the lookout for a simple walk and 180 degree turn around. Couldn't find a good one anywhere.

If you use Cinema 4d,  Dosch Design has motion that is precisely what you describe, and I think it's available for other programs as well. Also check out es3d.com's ambient moves which is tailored to Poser characters and works beautifully. They also take requests from vendors.

I guess it might depend on your project. I'm working on a 50+ minute project with very specific interaction between multiple characters--I don't think any mocap facility is going to give me that much mocap for under $2k (actually, if you know of one, let me know!) So, for me, a $2k system might be cost-effective. 

I have used bits of canned moves here and there, but overall they're generally not quite what I need, and I always have to edit the moves. It seems like having a real mocap solution that gave me the moves I wanted from the start would save me a lot of time as well.

--T

Quote - It seems like having a real mocap solution that gave me the moves I wanted from the start would save me a lot of time as well.

--T

 

I would love having a real mocap system of my own too, though I'm not sure this one would do multiple characters either. And it seems like there is always some editing of the figures involved, especially if you add speech to the characters.

Yes, the editing is a given--you're never going to get a perfect bvh right away, you'll always have to tweak, not to mention probably do a little cleaning, plus, as you say, adding speech (I use Mimic, a great program). That said, it would still help a lot to have the motions close to what I need to start. In any case, I hope yoshi continues to develop this solution--I'd like to see this succeed. It seems like mocap is one of the last parts of digital animation that's still horribly expensive, so it would be great to see a (relatively) inexpensive solution.--T

That $2000 price was for hardware which I have no control of.  That was about how much it costed when I setup my system.  I got everything brand new(tripods, USB 2.0 hubs, 15ft cables etc), obviously you don't need to pay for those if you already have them.  Also full body capture might work with 4 cameras, that will save another $500, but it will be less accurate, there will be more occlusions and markers will need to be manually identified more often.

I originally started this project when I was working on making a fight video game and manually creating motions was so tedious, it drove me almost crazy.  Even if nothing good comes out of this project, I will at least be able to continue my video game development ;-)

Quote - That $2000 price was for hardware which I have no control of.  That was about how much it costed when I setup my system.  I got everything brand new(tripods, USB 2.0 hubs, 15ft cables etc), obviously you don't need to pay for those if you already have them.  Also full body capture might work with 4 cameras, that will save another $500, but it will be less accurate, there will be more occlusions and markers will need to be manually identified more often.

I originally started this project when I was working on making a fight video game and manually creating motions was so tedious, it drove me almost crazy.  Even if nothing good comes out of this project, I will at least be able to continue my video game development ;-)

Well don't give up...I really feel that the Poser community would be all over this if it comes to fruition. I'm really looking forward to seeing how your final product works!

Yes, please stay with it! What you've shown us already looks great. If you get it to work to your satisfaction, you could probably sell it through Optitrack, since it would allow them to offer a complete system.--Tim

Agreed. Although It might be easier on the Poser user to not have to go through Optitrack exclusively to obtain the program. I could see an increase that might move it out of reach of a fair percentage of the DIY'ers. Plus most people want to choose things like tripods themselves.....

Yoshi, it looks like your program is working very well as it is. I saw the videos on your site and I'd say, even if the motion capture was limited (have to manually track some trackers when they move fast), it would still be an awesome piece of software to have. I'm really looking forward to seeing it released, if you choose to do so. You could release it under a beta program and then move to 1.0. I'd be very interested in beta testing this program or buying a copy...I'd get the cameras from optitrack tomorrow if your software was available for purchase or beta!

Any update news on this project....? Oh. And bump.

I've been trying to keep up on this too, so I spoke to someone at Optitrack the other day just to see what their take on software is, and it looks like they're pretty much counting on people like Yoshi to create the user interface(s) for the hardware. The person I spoke to seemed to allude to Yoshi's product as one that was in development. I emailed Yoshi after that, and he indicated that he hopes to have his app finished by the end of the year. Yoshi's site has an update dated 6/19.--Tim

Bump

Glad this was bumped since I didn't see it before. :)

This sounds like a good project and $2k isn't bad for the hardware. Any idea what price range you are looking at for the software?

Also, would you need an indoor studio or would this stuff work outside as well?

I like the fact that you would probably be able to take it anywhere to use and it is sort of in the price range for hobbyists as well.

I'll go out on a limb as say that outside is a very faint -maybe-. The optitrak cameras are infrared detectors; the cluster around the camera lens are hi out IR leds. Your two biggest problems would be consistent light intensities, and bleaching out of IR sources by ambient light (obviously this wouldn't work for anything other than an open [and probably roofed at the least] area. For something like being able to walk in the woods or around actual objects, you would need a lot more cameras and IR spotlights, or something like the Gypsy Gyro suit. Which is, shall we say, costly?). It might work on something like a solidly overcast day, but I would be highly doubtful about any other kind of exterior lighting conditions...

In addition to Dale's points, wind could move the camera from the calibrated positions.  And you don't want dirt or rain on your cameras/pc/usb hubs etc..  I have the cameras setup in my living room.  You don't need special studio or black background, as long as reflective stuff is out of camera's view.

Good point about the wind; there -are- wind resistant tripods (at least to up to 25 mph gusts IIRC), but they =cost=. And they are bulky. That's what's so jazz-worthy about this project. you're talking about 6 low cost digital cameras the size of your hand, 6 tripods that any camera could use. One or two USB 2 hubs and the cables to connect it all to a computer. Aside from the reflectors, all you would need would be said computer, and a supply of drapes to cover any reflective items, and you could take this anywhere. Dojo, dance studio, gym, ect (Although you would probably want to throw at least one standard video camera in the mix, just to have a record to keyframe finger and front of the foot motions from). Yoshi, does your program support multithreading?

yoshi please, you have think about the price of your software ? would be cool sell it here in marketplace ... with a good tutorial too about how setup your yoshi-mocap + optitrack cameras Best Regards; Andygraph

Sorry if I'm coming at this totally out of left field, but you might try a different camera setup. Directly overhead looking down, horizontally from the front, horizontally from the left and 45-degrees from the upper right should provide complete coverage of the space with fewer cameras, less data to process and a less expensive system.

Overall though this is a very exciting and promising endeavor.

Years ago I created a program for the Bronx Veterans Administration Hospital spinal cord injuries department and was heavily into Virtual Reality technology.  Data Gloves provided the ability to manipulate the virtual worlds for those that had some dextrous capabilites and IR tracking technology enabled quadraplegics to move a cursor across the screen just by moving their head while a "puff & sip" tube provided left & right mouse clicks so they could play chess, checkers or even a rudimentary game of tennis (for those of you who may remember REND386). I always felt that there was a therapeutic place for a mocap system to provide a form of visual feedback of limb position for those who couldn't derive that feedback from the nervous system directly. At that time, mocap technology like "Flock of Birds" was pretty slow, extremely  expensive and really glitchy. Primarily due to the extreme cost of the systems we were never able to implement that part of the program.

Perhaps your work will finally be able to bring this capability to some of those who have given so much to their country.

Keep up the good work and please keep me informed of your progress.

Regards,

cinesound@nyc.rr.com

 

<'))###<<

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

bump

And another bump and a new video (posted 0702)! I like that ability to select markers and get the XYZ graphs, yoshi. That is going to make initial cleanup a -lot- easier...

And another bump and a new video (posted 0702)! I like that ability to select markers and get the XYZ graphs, yoshi. That is going to make initial cleanup a -lot- easier...

Looking at the latest video, it looks like some of the markers are offset from the body.

Bug or Intentional? If intentional, why?

Looks to me as if it is either the difference between the size of the armature in the program vs the actual spatial positions of some of the markers, or a deliberate positioning to make it easier to find and identify the markers....

Very insteresting project! I wonder if wearing a suit with different color for each limb could replace the use of reflective markers. The software should calculate the "bounding box" for each colored area and from that calculate the orientation of the limb. ( I know, I'm making things too easy!!! ; )  )

P.S.

I suggest DAZ, as reseller.

My best wishes to yoshi-mocap
Fulvio