How's the best way to record two images coming through a single lens. One being a infrared and the other being visible. My plan was to use a beam splitter between the lens and the sensor to reflect the IR light into a second sensor while the visible light is transmitted to the other sensor. How would this mess with the focal point?

I'm just playing with the idea at the moment. Checking to see if it will work.

  • 1
    \$\begingroup\$ A beam splitter does not affect focus. \$\endgroup\$
    – Aganju
    Commented Oct 9, 2016 at 12:18
  • \$\begingroup\$ @Aganju - it does when it comes to IR. If you look at many older (film era) lenses you'll find they frequently feature an additional set of focus markers which are for use with IR film if nothing else. That's before you've started messing with the lens to film-plane distances needed to insert the splitter. \$\endgroup\$ Commented Oct 10, 2016 at 9:36
  • 1
    \$\begingroup\$ @JamesSnell no, the beam splitter does NOT affect focal length, although it may well affect coma if it's between the lens and the image plane. IR focal length differs from visible whether or not there's a splitter. \$\endgroup\$ Commented Oct 10, 2016 at 11:26
  • \$\begingroup\$ @Carl, you've misread. The splitter doesn't change the length but IR and visible light will need different distances from the split point for both to be in focus. \$\endgroup\$ Commented Oct 11, 2016 at 13:27
  • \$\begingroup\$ @JamesSnell I think you've misread. I was trying to point out that Aganju was correct and you were wrong in trying to correct him . The fact that IR needs a different focal length from visible is independent of the existence of a splitter in converging space. \$\endgroup\$ Commented Oct 11, 2016 at 13:44

2 Answers 2


As a theoretical problem it's something that is very much solved as it's how 3CCD video cameras often work (as the 3 indicates they have separate R, G, and B sensors.) Replicating something similar with SLR's should be achievable though it may take some fettling to get right, it would definitely be a homebrew type project.

Splitting visible light from IR may introduce some challenges. You'd need a body modified to handle IR in addition to a standard SLR body. To maintain focus the sensors would also need to be slightly different distances from the split point so they're unlikely to line-up exactly even if the two bodies are the same model.

Then it's a case of syncing up the releases which can be done with a remote and that's it, you're done.

  • \$\begingroup\$ What happens if I need to increase the distance from the back of the lens to the sensor. Say take it from the default distance and add 30mm so I can fit in the splitter. I'm I right in thinking that Increasing the distance from the back of the lens will apply a zoom to captured image? Is there an attachment I can get for the back of the lens that bends the light so I can increase this distance without applying a zoom? \$\endgroup\$
    – colyton
    Commented Oct 13, 2016 at 8:55
  • \$\begingroup\$ There would definitely be a zoom effect (a crop technically) and a corresponding loss of light with it. Your other option would be to split before the lens. A half-silvered mirror ought to see you right ahead of two near-identical lenses and a couple of primes wouldn't be excessive compared to the cost of modifying a DSLR for IR. You'd probably get better output that way. \$\endgroup\$ Commented Oct 13, 2016 at 10:33
  • \$\begingroup\$ Do you know of a calculation I could use to compute the loss of light? I'm going to be using two basler computer vision camera not dslr. So I not worried about the cost of modifying a dslr to record ir. Does half-silvered reflect ir? Yea the duel lens like you mentioned might be the idea solution. Just a little bit more difficult if I'm not using prime lens. Thanks for the knowledge James \$\endgroup\$
    – colyton
    Commented Oct 14, 2016 at 2:06
  • \$\begingroup\$ I don't know about the mirror and IR, would be worth checking with the physics stack and see what they say. \$\endgroup\$ Commented Oct 14, 2016 at 20:19

You'll be a lot better off if you can afford two separate lenses & sensors, as then each path (IR, vis) can be focussed separately. You may be able to adjust the back distance from the lens to each sensor, but it's never a good idea to put optics in the converging portion of the path (between lens and sensor). A tilted flat plate in the cone will cause coma in the image, for example.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.