Hello. I have a proposed simulation build that I am very interested in getting some input with. I have posted all of the details here:
Minus Reality Blog http://www.minus-reality.com/?p=18
Cost-Effective Stereoscopic 6DOF VR Simulation Build
-
Seth Steiling
- Posts: 1365
- Joined: Fri Jun 27, 2008 11:29 am
- Location: Corvallis, Oregon
Re: Cost-Effective Stereoscopic 6DOF VR Simulation Build
I'm not sure that I'm understanding your plans correctly, but it sounds like you're looking to create a tracked volume that is between 50 x 50 and 100 x 100 feet? If so, you're not going to accomplish that with 3 of our USB cameras. Even with 24 V100:R2 cameras, the maximum volume is around 15 x 15 ft, with the possibility of jumping up to around 17 x 17 ft under best circumstances.
Marketing Manager
TrackIR | OptiTrack
TrackIR | OptiTrack
Re: Cost-Effective Stereoscopic 6DOF VR Simulation Build
Ok, maybe I'm a little confused. I have been told that 25 feet is the max tracking distance with these cameras using active LEDs. Also I am not talking about full body motion capture but head and maybe also hand tracking. Would these thing change the maximum trackable volume at all?
Re: Cost-Effective Stereoscopic 6DOF VR Simulation Build
Ok, there is another confusing thing going on here. When people on this forum describe their multi-camera systems and the requirements, it seems that they are referring to full-body motion capture. There are certain assumptions here:
1) A full body must be viewed by multiple angles to prevent occlusion
2) The cameras must surround the user to provide 6DOF data from various angles
Tell me if I'm wrong here. But it seems that you can get 6DOF head tracking data from only one direction (top down)..of course this assumes that the user stays in a relatively upright position.
This said, couldn't you use a larger area of cameras pointing downards from aabove and stretch that trackable area much further than that which would be required to track a full body since occlusion is no longer as much of a factor?
Under this configuration the closest nearby individual cameras are never more than maybe 12ish feet from the user's head So the array of cameras now form a flat plane pointing downards instead of a cube.
Is this feasible? Am I missing something?
1) A full body must be viewed by multiple angles to prevent occlusion
2) The cameras must surround the user to provide 6DOF data from various angles
Tell me if I'm wrong here. But it seems that you can get 6DOF head tracking data from only one direction (top down)..of course this assumes that the user stays in a relatively upright position.
This said, couldn't you use a larger area of cameras pointing downards from aabove and stretch that trackable area much further than that which would be required to track a full body since occlusion is no longer as much of a factor?
Under this configuration the closest nearby individual cameras are never more than maybe 12ish feet from the user's head So the array of cameras now form a flat plane pointing downards instead of a cube.
Is this feasible? Am I missing something?
-
Seth Steiling
- Posts: 1365
- Joined: Fri Jun 27, 2008 11:29 am
- Location: Corvallis, Oregon
Re: Cost-Effective Stereoscopic 6DOF VR Simulation Build
Yes, it is possible to create a capture volume comprised entirely of ceiling mounted cameras facing downward. However, in order to calibrate such a system you would need moderate field of view overlap between neighbor cameras for both calibration and tracking.
So, yes, tracking a few simple rigid bodies is definitely simpler than a full body. And that would likely enable you to create a camera arrangement that focuses more on maximizing the volume size than overlapping camera view. But you still need to have camera overlap.
The 25 foot tracking range means that an individual camera could see an individual LED, in 2D, up to that distance away from that camera. It doesn't translate to 25 feet of capture volume, or 25 feet of camera coverage.
You can think of each camera's field of view as a cone radiating out from each camera in 3D space to the max tracking range (25 feet for R2). A marker needs to be detected within at least two of these cones in order to be triangulated.
For USB cameras there will also be some cable length limitations due to the number of active extension cables that can be used between the host PC and USB devices. Our Ethernet based cameras do not have this limitation.
For conventional side mounted rigs, a major generalization of a volume's baseline capability (because it really varies based on numerous factors) you can assume (if tracking a single rigid body with V100:R2 cameras and wide angle lenses):
- 3 cameras max out at around 3' x 3' to 5' x 5'
- 6 cameras max out at around 8' x 8' to 10' x 10'
- 8-12 cameras max out at around 12' x 12' to 14' x 14'
- it might be possible to track volumes up to 17' x 17' or so with lots of cameras, a good calibration, creative camera aiming, active markers, etc.
So, yes, tracking a few simple rigid bodies is definitely simpler than a full body. And that would likely enable you to create a camera arrangement that focuses more on maximizing the volume size than overlapping camera view. But you still need to have camera overlap.
The 25 foot tracking range means that an individual camera could see an individual LED, in 2D, up to that distance away from that camera. It doesn't translate to 25 feet of capture volume, or 25 feet of camera coverage.
You can think of each camera's field of view as a cone radiating out from each camera in 3D space to the max tracking range (25 feet for R2). A marker needs to be detected within at least two of these cones in order to be triangulated.
For USB cameras there will also be some cable length limitations due to the number of active extension cables that can be used between the host PC and USB devices. Our Ethernet based cameras do not have this limitation.
For conventional side mounted rigs, a major generalization of a volume's baseline capability (because it really varies based on numerous factors) you can assume (if tracking a single rigid body with V100:R2 cameras and wide angle lenses):
- 3 cameras max out at around 3' x 3' to 5' x 5'
- 6 cameras max out at around 8' x 8' to 10' x 10'
- 8-12 cameras max out at around 12' x 12' to 14' x 14'
- it might be possible to track volumes up to 17' x 17' or so with lots of cameras, a good calibration, creative camera aiming, active markers, etc.
Marketing Manager
TrackIR | OptiTrack
TrackIR | OptiTrack
Re: Cost-Effective Stereoscopic 6DOF VR Simulation Build
[quote=NaturalPoint - Seth]
- 3 cameras max out at around 3' x 3' to 5' x 5'
- 6 cameras max out at around 8' x 8' to 10' x 10'
- 8-12 cameras max out at around 12' x 12' to 14' x 14'
- it might be possible to track volumes up to 17' x 17' or so with lots of cameras, a good calibration, creative camera aiming, active markers, etc. [/quote]
Thank you so much for clarifying these things for me Seth. Could you possibly give me the same numbers for a s250e based system?
- 3 cameras max out at around 3' x 3' to 5' x 5'
- 6 cameras max out at around 8' x 8' to 10' x 10'
- 8-12 cameras max out at around 12' x 12' to 14' x 14'
- it might be possible to track volumes up to 17' x 17' or so with lots of cameras, a good calibration, creative camera aiming, active markers, etc. [/quote]
Thank you so much for clarifying these things for me Seth. Could you possibly give me the same numbers for a s250e based system?