What is the optimal capture area for a 12-camera system? I couldn't seem to find an official recommendation for this on the site although I have heard mention of 20 square feet? What would the recommended lenght, height and width of the capture area be to ensure optimum results?
I am looking at renting new premises and would naturally be looking for a suitable area to house the OptiTrack system (which I have yet to order!).
Thanks in advance.
Optimal Capture Area
Re: Optimal Capture Area
Good question. We are working on some new layout diagrams now, for optimal placement of 12, 18 and 24 cameras.
There really are two questions here, one is the maximum or ideal capture volume you can create with 12 cameras and the other is the camera placement or mounting for that volume.
For the capture volume size, we have had good luck with a 15'X20' space, cameras mounted on a truss work 9' high. This gave about a 7'X10' capture volume. You could easily extend this to a 20' square space and get a 10'x10' capture volume. You could certainly try 25' square, but I would recommend that for 16-18 cameras to get better coverage with that size space.
For camera mounting, you really have a lot of options. In general, we can talk about where the cameras are located and how you mount them, or get them there is up to you. Mounting can be; wall mounts, tripods, autopoles (great for tight volumes), truss, or pipe systems (speed rail).
For tripods, I would use 6 of them, with 2 cameras mounted on each one, one at 8'+ and the other at 4-5', like our 6 camera diagram, just with 2 cameras per tripod. This can take up space, as the tripods legs are wide, autopoles work well for this style of mounting.
For truss, or anything other than tripods, and what we use here, is 2 cameras in each corner, one 8'+, the other 4-5' (like the 8 camera diagram) and then fill in the spaces between the 4 corners with one camera per side, so 4 more total.
You might also consider rotating the lower cameras 90 degrees, so they are in the portrait position, this tends to give better coverage.
Hope this helps, hard to describe with text, so we will work on the drawings right away.
Also, thanks for considering an OptiTrack system!
There really are two questions here, one is the maximum or ideal capture volume you can create with 12 cameras and the other is the camera placement or mounting for that volume.
For the capture volume size, we have had good luck with a 15'X20' space, cameras mounted on a truss work 9' high. This gave about a 7'X10' capture volume. You could easily extend this to a 20' square space and get a 10'x10' capture volume. You could certainly try 25' square, but I would recommend that for 16-18 cameras to get better coverage with that size space.
For camera mounting, you really have a lot of options. In general, we can talk about where the cameras are located and how you mount them, or get them there is up to you. Mounting can be; wall mounts, tripods, autopoles (great for tight volumes), truss, or pipe systems (speed rail).
For tripods, I would use 6 of them, with 2 cameras mounted on each one, one at 8'+ and the other at 4-5', like our 6 camera diagram, just with 2 cameras per tripod. This can take up space, as the tripods legs are wide, autopoles work well for this style of mounting.
For truss, or anything other than tripods, and what we use here, is 2 cameras in each corner, one 8'+, the other 4-5' (like the 8 camera diagram) and then fill in the spaces between the 4 corners with one camera per side, so 4 more total.
You might also consider rotating the lower cameras 90 degrees, so they are in the portrait position, this tends to give better coverage.
Hope this helps, hard to describe with text, so we will work on the drawings right away.
Also, thanks for considering an OptiTrack system!
Re: Optimal Capture Area
At some point I'd really like to try a more traditional overhanging grid setup with the cameras. In this scenario, you hang the cameras overhead in rows with different orientations. They look at the volume at a steep angle, say between 45 and 80 degrees. At that point, lensing, camera height and resolution play off of one another and determine placement further.
Making the volume larger at that point, is just a matter of adding more rows of cameras. So, rather than being on the edge and looking in, cameras can be shooting from just above any part of the volume. They also very rarely see each other as they're all angled down.
As markers move across the larger volume, cameras essentially hand off the marker (not really any different than when someone turns away from some cameras and in turn reveals to others... just positional rather than rotational). The advantage for optitrack cameras in this case, I think, would be that the issue of marker distance is dealt with rather nicely. The maximum marker distance pretty much just limits how high above the volume you can mount your cameras (and the current limitations on the v100 and its sensor would make for a totally reasonable height). The edge-in configurations will eventually result in a doughnut problem where the center of the volume can not be seen due to its distance from the edge where the cameras are.
With 8 cameras, you can't really set up a grid pattern. But when you start getting into the 24-48 range, I think it starts to become the preferred solution, if its possible.
Arena would need to shed its current issues when it comes to seeing the calibration wand in a subset of cameras rather than all the cameras, to accomplish this, but I know you guys are working on that and I have faith
Making the volume larger at that point, is just a matter of adding more rows of cameras. So, rather than being on the edge and looking in, cameras can be shooting from just above any part of the volume. They also very rarely see each other as they're all angled down.
As markers move across the larger volume, cameras essentially hand off the marker (not really any different than when someone turns away from some cameras and in turn reveals to others... just positional rather than rotational). The advantage for optitrack cameras in this case, I think, would be that the issue of marker distance is dealt with rather nicely. The maximum marker distance pretty much just limits how high above the volume you can mount your cameras (and the current limitations on the v100 and its sensor would make for a totally reasonable height). The edge-in configurations will eventually result in a doughnut problem where the center of the volume can not be seen due to its distance from the edge where the cameras are.
With 8 cameras, you can't really set up a grid pattern. But when you start getting into the 24-48 range, I think it starts to become the preferred solution, if its possible.
Arena would need to shed its current issues when it comes to seeing the calibration wand in a subset of cameras rather than all the cameras, to accomplish this, but I know you guys are working on that and I have faith
Re: Optimal Capture Area
Brad:
Good idea. We could try out the grid idea here, with 24 cameras mounted on the Speed Rail we discussed. That should be inexpensive and easy to prototype. Would you be up for putting a quick 3D rendering of how you would position the cameras online?
Good idea. We could try out the grid idea here, with 24 cameras mounted on the Speed Rail we discussed. That should be inexpensive and easy to prototype. Would you be up for putting a quick 3D rendering of how you would position the cameras online?
Re: Optimal Capture Area
Sure. I'm flying to texas tomorrow and starting a new job. So it may be a bit before I have a chance to put it together. But I'll make it a priority when I have free time.
Re: Optimal Capture Area
I've uploaded two files to my website.
http://www.fie.us/tempdl/cams24_45fov.ma
http://www.fie.us/tempdl/cams24_60fov.ma
They are untested examples of the kind of camera setup I've seen as the more conventional approach to large volume captures.
As you can see, the lense choice has a little to do with how far your grid is off the ground. The last one of these I dealt with was hanging from a lighting grid in a sound stage. So it was a good 40 feet in the air. They compensated by going with longer lenses. If however, you had a lower ceiling, shorter lenses become appropriate so as not to waste resolution.
http://www.fie.us/tempdl/cams24_45fov.ma
http://www.fie.us/tempdl/cams24_60fov.ma
They are untested examples of the kind of camera setup I've seen as the more conventional approach to large volume captures.
As you can see, the lense choice has a little to do with how far your grid is off the ground. The last one of these I dealt with was hanging from a lighting grid in a sound stage. So it was a good 40 feet in the air. They compensated by going with longer lenses. If however, you had a lower ceiling, shorter lenses become appropriate so as not to waste resolution.