We have an OptiTrack FLEX:V100 which we've been using to track markers. The markers are made with black paper taped onto an IR-reflective material and using an LED intensity of 12 (max) and exposure of 2 (of 399) the markers showed up beautifully, to the point where nothing but the markers were seen.
Today we decided to switch from the FLEX:V100 to a pair of SLIM:V100 cameras. With all the cameras on at once, and side-by-side, the FLEX:V100 produces a perfect result with only the markers seen while the SLIM:V100 cameras can barely see anything (mostly black). The FLEX:V100 is producing IR light for the SLIM:V100, along with an array of IR LEDs. So, my question is what could we do to make the SLIM:V100 cameras produce a similar result to what we experienced from the FLEX:V100?
What we tried:
Lenses: Currently both the SLIM:V100 cameras have wide-angle lenses. Using a normal lens does not change the result, however.
Filters: We tried taking out the filter from one of the SLIM:V100s. The interesting thing is that the camera with the filter produced a brighter image than the one without the filter.
Exposure: With exposures around 50 and more, the markers are visible, but so are much of the surroundings, leading to unstable marker detection.
Any help would be greatly appreciated.
Thanks!
FLEX:V100 Results vs. SLIM:V100 Results
Re: FLEX:V100 Results vs. SLIM:V100 Results
There are two potential issues :
1) Its likely that the exposure/strobe timing of the FLEX:V100 is not synchronized with the SLIM:V100s. Since the exposure/strobe window is so short (50 = 1/1,000th second, 2 = 1/25,000th second), its easy for them to not align temporally.
2) Retro-reflective material only has good reflective performance when the light source originates as close to the camera lens as possible. So the SLIMs would only be able to pick reflections of the FLEX's IR LEDs off the reflective material if they were right behind the FLEX (which would makes it hard for them to see the target). The amount of reflected light falls off dramatically as you move off axis.
There are two ways to resolve the timing issue (this won't solve the retro-reflective condition though) :
a) Wire up the SLIMs to use the FLEX's sync-output, then they will automatically align to its exposure/strobe timing.
b) Switch the FLEX from strobe mode to constant illumination mode (I think a value of 1-7 in current SDK builds). This would probably require longer exposure times (100+) on all cameras to pick up the same amount of light.
To solve the retro-reflective issue, you'd probably need to provide each SLIM with its own light source. Some customers have done this by placing large lights directly behind each camera.
1) Its likely that the exposure/strobe timing of the FLEX:V100 is not synchronized with the SLIM:V100s. Since the exposure/strobe window is so short (50 = 1/1,000th second, 2 = 1/25,000th second), its easy for them to not align temporally.
2) Retro-reflective material only has good reflective performance when the light source originates as close to the camera lens as possible. So the SLIMs would only be able to pick reflections of the FLEX's IR LEDs off the reflective material if they were right behind the FLEX (which would makes it hard for them to see the target). The amount of reflected light falls off dramatically as you move off axis.
There are two ways to resolve the timing issue (this won't solve the retro-reflective condition though) :
a) Wire up the SLIMs to use the FLEX's sync-output, then they will automatically align to its exposure/strobe timing.
b) Switch the FLEX from strobe mode to constant illumination mode (I think a value of 1-7 in current SDK builds). This would probably require longer exposure times (100+) on all cameras to pick up the same amount of light.
To solve the retro-reflective issue, you'd probably need to provide each SLIM with its own light source. Some customers have done this by placing large lights directly behind each camera.
Re: FLEX:V100 Results vs. SLIM:V100 Results
Thank you for the reply, I will look into getting a new light source.
Cheers,
Andrei
Cheers,
Andrei