We are interested using using structured (lines or other patterns or combinations thereof) light pattern in multi-camera setups for surface reconstruction and tracking. In particular we would like to experiment with light in the NIR range. We would be interested in using a few (or even several) wavelength bands.
Does the DLP® Pico™ Projector allow for substituting the red, blue and green LEDs with NIR LEDs?
Can you point me to other solutions in finding a versatile NIR video projector
Rasmus Larsen, Technical University of Denmark
Rasmus,
Let me welcome you to the forums! It is great to have you .
I hope that the forum will be of help with your questions.
We are looking to see how practical it would be and whether the lens coatings will permit it. I will try to get back to you soon.
I can't say specifically about the Pico kit, but I would be cautious.
Most optics come with coatings that significantly attenuate IR, even NIR, unless they have been specifically designed for IR operation. That said, the DMD's coating does pass NIR at 90% range of efficiency, as long as you aren't going out too much beyond 850nm or so. If you can live with very low throughput, you might be able to make do.
Another issue is that RGB optics (including the Pico) will not correct for IR wavelengths. Depending on the specifics of the prescription, this could lead to significant loss of projected MTF, astigmatism, etc. -- all of which tend to ruin structured lighting performance.
I wish there were a more positive answer, but my suggestion would be to test your application in visible light. If it works for your needs, our DMDs will do just fine in NIR and a custom NIR-only optical design would be very feasible (for a company that has done DLP optics before) and could certainly achieve the same performance as an RGB design.
for the Pico kit, I think you'd have to bypass the LED colour-combining optics completely, as these appear to be rather narrowband, and so may not let nay NIR through at all - I got almost no visible transmission from a blu-ray laser through the blue filter, but might be worth a try blasting a load of NIR through the red LED port before tearing further into the optical block.
Right, the LEDs are blasting through color dichroic mirrors to combine them. Since all you'd need for NIR is a single "color", you'd just remove all the color LEDs and put the IR LED in the nearest position, replacing the dichroic with a simple mirror. In theory anyway.
Another possibility is to replace the dichroics with narrowband filters that pass different small bandwidths of NIR spectrum. Then replace the RGB LED's with NIR LED's that match these bandwidths, if there is such a thing, so that the full combined bandwidth of all 3 LED's gets to the DMD.
Of course there is still the problem of how efficient the optics would be at NIR since they were optimized for focus and performance at visible wavelengths. Not to mention the bandpass of the optical materials themselves, which includes acrylic optics.
Mike Davis