Red UFOs during long exposures with Canon 24-104mm f/4L lens

If you thought the Canon 5D Mark II’s black dot problem was obscure, here is an even stranger defect in some Canon cameras. The issue (red flare on multi-minute exposures) was originally attributed to the a problem with an LED on the Canon 5D Mark II – but this proved to be wrong. In January 2009 a DPReview forum user known as “darksbrother” posted:

Ok, I’ve got a conundrum. I need to find a way to disable the blinking red LED to the right of the wheel on the 5D Mark II. I’m doing long night exposures and that LED keeps creating a red streak on sufficiently long enough exposures.

Firstly, apparently that LED on the back of the camera is on during long exposures. Which is in itself surprising, and lead to some discussion because users were not aware of this. Secondly, the original poster cared about that red LED only because it seemed to cause a red spot in the lower part of dark images during multi-minute exposures. In other words, the poster thought that light from the LED somehow leaking via the internals of the camera and managed to reach the exposed CMOS sensor. Within a few days, it was announced on the DPReview forum by its_RKM that this was a problem with the Canon 24-104mm f/4L lens rather than with the camera body: infrared light from an infrared LED inside the lens (used to detect motion for the USM mechanism) leaked into the camera at certain zoom ranges (notably 24 mm). The Canon 5D Mark II sensor understandably registered this infrared light as red light – despite the infrared absorbing filter that is the light needs to pass through before reaching the sensor. If you expose long enough (e.g. 15 minutes) and set the camera to automatically subtract a 2nd image (also 15 minutes!) taken with the shutter closed, the resulting noise-reduced image will show red “flare” as shown in the image

Red UFO image posted by RKM to DPReview (15 minutes, 24 mm, f/4)

Image posted at DPReview by its_RKM on January 31st (15 minutes, 24 mm, f/4, no external light)

On March 31st 2009, the original poster summarized the forum thread as

Thanks to extensive testing by RKM, we can in fact conclude it’s the lens. I have tried the test with the body cap and have gotten no red streak. What was concluded was that it was an internal infrared LED within the lens which helps calibrate focus/read information on the position of the lens that was leaking through into the sensor area causing the red streak.

but Canon technical support is so far unwilling/unable to confirm the situation. Technically interesting questions are:

  1. Does this show up with all SLR cameras using the same lens? So far, it has been seen on a 5D and 5D Mark II. The problem may be less visible on an APS-C sensor as the worst parts would be cropped off anyway.
  2. Does the problem still occur if you disable autofocus? Manual focus may turn the LED off (although that’s less likely).
  3. How much infrared does the IR cut filter actually block (this link suggests at least 5 stops for 800 nm light)? An ancient Canon 10D might be sensitive to it because it said to have a weak IR cut filter (on the other hand, the sensor is noisy by current standards anyway).
  4. Are there other Canon USM lenses with the same problem? Most high-end lenses have a similar USM focus technology.
This entry was posted in Photo Technology and tagged , . Bookmark the permalink.

9 Responses to Red UFOs during long exposures with Canon 24-104mm f/4L lens

  1. Craig Gaskell says:

    I have this problem too and it makes a mess of Astro photography. Do you know if there is a fix?


    • pvdhamer says:


      The original thread at DPReview (link in posting) doesn’t contain a fix and the thread is no longer active. I am incidentallly not into astrophotography myself (I was just curious).

      I vaguely remember somebody in that tread tried leaving their lens “unclicked” in the mount to prove the phenomenon: by leaving the lens rotated in the mount, the electrical contacts shouldn’t be aligned, and the infrared source inside the lens can’t be powered. This implies shooting at maximum aperture. Similarly, you can tape off the contacts with thin tape instead of taking mechanical risks.

      Just to be sure: what body, what lens, what exposure times are we talking about?


  2. Robert says:

    Good post – wish I had found it earlier. I noticed a similar problem with an astronomy-converted Canon t2i and the 24-105mm L. Without the IR filter, the problem shows up as a purple fog with increasing brightness towards one corner of the image. I spent several hours with the camera apart masking off any light coming from the red activity LED in the right bottom corner of the back of the camera – all to no avail. So I can confirm that the IR filter on the sensor does block a significant amount of this (I can get visible artifacts with far less exposure time than the example above – for instance with the above camera/lens combination with exposures in the 10-15 second range at ISO 800).

    Craig – Two fixes: 1) use primes – Both my 24-105mm f/4 L and my 17-40mm f/4 L suffer from this issue, but the 17-40mm to a lesser degree. I have no issue with my 50mm f/1.4 or my 14mm f/2.8L. 2) As Peter suggested (and I’ve found it works), leave the lens partially engaged “unclicked” so that the electrical contacts don’t engage.


  3. Brian Allan says:

    This is problem much mentioned on astro-imaging forums w/ the 5D, in most cases with modified (full spectrum) cameras where there is no longer an IR cut filter over the sensor. However, I can understand it also being a problem with standard cameras since the IR cut filter does pass some IR w/ a long enough exposure…

    • pvdhamer says:


      I don’t do astrophotography (and was merely passing on the info in this case). So it is useful to hear that this problem shows up much more clearly when the IR filter has been removed. Thanks.


  4. It is possible to use the ‘twisted lens’ trick AND still select the working aperture.

    If you use AV or Manual mode select the aperture you want to use. Then press the Depth of Field preview button, with the DoF button pressed release the lens from the camera body and twist a few degrees to uncouple the electronics. If you watch the LCD display as you twist the lens the aperture will change to f00 when the lens is twisted far enough

  5. Ted Mickle says:

    I have a Canon 7D that was modified for astronomy and am seeing blue streaks appearing, even with test shots (5-10 seconds) in a dark room — using the 28-135mm image stabiler zoom lens with the lens cover on.

    I sent the camera back to the vendor who performed the modifications, but they cannot reproduce the problem.

    Could the problem be in the lens itself, even with the autofocus feature turned off?

    • pvdhamer says:

      Ted, I don’t actually know more about this than I wrote down as I didn’t discover the phenomenon or even bother to confirm it (it had been confirmed enough). So I can only help with debugging suggestions:
      1) yes, the problem could certainly be in the lens. If the 24-105mm has it, it is likely that some other Canon lenses also have the problem.
      2) to determine whether it is lens or body, you can replace the lens with a plastic body cap. Or use the tricks whereby the lens is slightly rotated to decouple the electrical contacts.
      3) I cannot explain why you would get “blue” rather than red. The 24-105mm certainly uses infrared, and that gets into the red channel. It sounds likely that other lenses also use infrared (cause that’s what usually used for optocouplers and coding disks). So blue sounds suspicious. I assume your modified 7D still handles normal color photos correctly (e.g. no shifted Color Filter Array)?4) keep in mind that there can be light leaking into the camera via the viewfinder. That might explain the blue. Some camera straps come with pastic sliders to cover the eyepiece to avoid light leakage.


  6. Bwana says:

    Many other lenses (from various manufacturers) display this internal IR light pollution problem. If you join the DSLRModifications forum, there is listing of lenses confirmed to have the problem and much discussion about the problem and potential fixes…

Leave a Reply

Your email address will not be published. Required fields are marked *