Are there any advantages to active (IR leds) markers vs. passive markers? Are they tracked better, more accurate?
Will
Markers..
Re: Markers..
they can be seen from a much longer distance usually. they can be seen under more complex lighting conditions. In theory you can purposely cycle them on and off to help with IDing and labeling them. Either through pulse codes or simple sequencing.
the prototypical example is the Vampire brides from Van Helsing. ILM shot them in full makeup in blue suits on a soundstage with full film style lighting and needed full body mocap in sync.
Therefore, they used active markers as there was a lot of infrared light coming off the film lights. They needed their markers to rise above the chatter and spill.
It is probably worth noting that ILM eventually moved to their newer mocap style as seen in Pirates of the Caribbean after their Van Helsing experience, rather than moving forward with the active marker approach. I'm told their original markers tended to heat up quite a bit. But knowing what I do about IR LEDs and mocap, I suspect thats the result of a rushed hardware design.
Its unclear how much of their newer pipeline is automatic and how much is manual. However, their suit and marker design clearly imply algorithms more commonly associated with the larger field of computer vision rather than the subset of computer vision usually applied to optical mocap in production. Their 'circle in checkerboard' patterns don't appear to simply be for increased contrast. There's something more going on there.
Anyhow, I've played with the LED's on the NP site. They're pretty impressive. I've not put together a full active marker set just yet. I'm still trying to optimize the reflective performance of the system. However, I will probably move on the active markers within a few months I suspect.
the prototypical example is the Vampire brides from Van Helsing. ILM shot them in full makeup in blue suits on a soundstage with full film style lighting and needed full body mocap in sync.
Therefore, they used active markers as there was a lot of infrared light coming off the film lights. They needed their markers to rise above the chatter and spill.
It is probably worth noting that ILM eventually moved to their newer mocap style as seen in Pirates of the Caribbean after their Van Helsing experience, rather than moving forward with the active marker approach. I'm told their original markers tended to heat up quite a bit. But knowing what I do about IR LEDs and mocap, I suspect thats the result of a rushed hardware design.
Its unclear how much of their newer pipeline is automatic and how much is manual. However, their suit and marker design clearly imply algorithms more commonly associated with the larger field of computer vision rather than the subset of computer vision usually applied to optical mocap in production. Their 'circle in checkerboard' patterns don't appear to simply be for increased contrast. There's something more going on there.
Anyhow, I've played with the LED's on the NP site. They're pretty impressive. I've not put together a full active marker set just yet. I'm still trying to optimize the reflective performance of the system. However, I will probably move on the active markers within a few months I suspect.
Re: Markers..
Bred,
Thanks for the info. When you mean 'seen from a longer distance', is it possible to extend the 10 x 10, 6 camera set up with IR markers? I work in the film business doing make up effects and actually worked on "Van Helsing". http://www.imdb.com/name/nm0400235/ I am also working on "Watchman", where they are employing the checkboard pattern for mocap.
Anyways on Van Helsing if I remember correctly they had LEDs in a transluscent ball. I think I will build a suit with these LED's when I get my system and see if I get better tracking results.
As a makeup artist, I've kinda starting teaching myself CG animation from about 1997. I'd like to employ mocap to some characters I created in Zbrush for a trailer in an all CG project. Hopefully I will be able to mocap my motions with Arena and stop using mocap from the net.
Thanks,
Will
Thanks for the info. When you mean 'seen from a longer distance', is it possible to extend the 10 x 10, 6 camera set up with IR markers? I work in the film business doing make up effects and actually worked on "Van Helsing". http://www.imdb.com/name/nm0400235/ I am also working on "Watchman", where they are employing the checkboard pattern for mocap.
Anyways on Van Helsing if I remember correctly they had LEDs in a transluscent ball. I think I will build a suit with these LED's when I get my system and see if I get better tracking results.
As a makeup artist, I've kinda starting teaching myself CG animation from about 1997. I'd like to employ mocap to some characters I created in Zbrush for a trailer in an all CG project. Hopefully I will be able to mocap my motions with Arena and stop using mocap from the net.
Thanks,
Will
Re: Markers..
Ah, Watchmen. I've worked with the VFX Sup, DJ on a couple of projects this past year. He's a good guy.
Regarding the translucent ball, Jim and I have both messed around with IR LEDs for this kind of stuff before. The translucent domes tend not to work so hot IMHO. They usually cut the light too much. The wide throw LEDs that Jim has in the natural point store are pretty good on their own due to their wide angle of projection. If you're looking for diffusion, I find that rather than putting material in front of the LED, carefully sanding the surface of the plastic dome tends to work a lot better. Jim may have some tricks of his own?
Yea, the intent is to get a 10 x 10 human scale volume larger. But understand, that you may run up against the resolution issue. You may make it larger, but its internal resolution may be too low to be useful for human capture without adding cameras. I've not tested much larger than 15x15 just yet so I'm not quite sure of the actual performance.
I'm working on a few other ways to augment the capture distance of the cameras. Its a little too soon to tell if I've been successful. However, one thing I have confirmed:
halogen bulbs (I'm using a 60watt projector bulb from home depot) put out a lot of infrared light. To me, it looks like a single 60watt bulb puts out about as much light as the LEDs on the camera itself. If you position the bulb behind or just next to the camera, the retro reflective material does bounce the light back to the lense. I would think a heat lamp might be even better, but they make so much heat, they may be impractical for actual use (don't want that thing too close to the camera).
The jury is still out on the usefulness of this trick.
Regarding the translucent ball, Jim and I have both messed around with IR LEDs for this kind of stuff before. The translucent domes tend not to work so hot IMHO. They usually cut the light too much. The wide throw LEDs that Jim has in the natural point store are pretty good on their own due to their wide angle of projection. If you're looking for diffusion, I find that rather than putting material in front of the LED, carefully sanding the surface of the plastic dome tends to work a lot better. Jim may have some tricks of his own?
Yea, the intent is to get a 10 x 10 human scale volume larger. But understand, that you may run up against the resolution issue. You may make it larger, but its internal resolution may be too low to be useful for human capture without adding cameras. I've not tested much larger than 15x15 just yet so I'm not quite sure of the actual performance.
I'm working on a few other ways to augment the capture distance of the cameras. Its a little too soon to tell if I've been successful. However, one thing I have confirmed:
halogen bulbs (I'm using a 60watt projector bulb from home depot) put out a lot of infrared light. To me, it looks like a single 60watt bulb puts out about as much light as the LEDs on the camera itself. If you position the bulb behind or just next to the camera, the retro reflective material does bounce the light back to the lense. I would think a heat lamp might be even better, but they make so much heat, they may be impractical for actual use (don't want that thing too close to the camera).
The jury is still out on the usefulness of this trick.
Re: Markers..
Brad,
Yeah DJ is a good guy. They have alot on their plate with this one.
I assumed the IR balls were to give a larger target to track. With the resolution of the camera I thought it would be a problem, but since they are wide angle, its easier for the camera to pick up..so the balls are not needed. Gotcha!
Keep us informed as to your progress on the bulbs.
Will
Yeah DJ is a good guy. They have alot on their plate with this one.
I assumed the IR balls were to give a larger target to track. With the resolution of the camera I thought it would be a problem, but since they are wide angle, its easier for the camera to pick up..so the balls are not needed. Gotcha!
Keep us informed as to your progress on the bulbs.
Will
Re: Markers..
Interesting stuff. I agree with Brad, I like the raw domed LEDs that we sell (of course), and you can sand the epoxy dome if you want to.
We are focusing supporting more cameras per PC, as a way to increase capture volume. The V100 has a useful illumination range of about 25' for a 5/8" marker, maybe 20' being conservative. So, adding more cameras is good because it increases the resolution and helps spread them out correctly.
60 watt bulbs do put out a ton of IR, that is a useful trick, but it can be not a lot of fun to be performing under visible light.
We are focusing supporting more cameras per PC, as a way to increase capture volume. The V100 has a useful illumination range of about 25' for a 5/8" marker, maybe 20' being conservative. So, adding more cameras is good because it increases the resolution and helps spread them out correctly.
60 watt bulbs do put out a ton of IR, that is a useful trick, but it can be not a lot of fun to be performing under visible light.
Re: Markers..
hehe. true. but thats what visible light filters are for