1D X can AF at f/8!

Discussion in 'Canon EOS' started by yakim_peled|1, Sep 4, 2012.

  1. I kid you not. Look here. What I can't understand is why the 5D3 could not AF in such conditions.
    Happy shooting,
    Yakim.
     
  2. It's odd. It does it only with the Kenko TC and that in itself is not surprising. I get excellent AF with a Tamron 1.4X TC and a Canon EF 500/4.5L (effective f6.3), whereas AF shuts off with the Canon 1.4x TC (5D and 7D).
    What's odd is the difference between the 5D MkIII and the 1D X since they both use the same AF sensors. The 5D MkIII seems to be recognizing the Kenko as a TC, while the 1D-X doesn't seem to know it's there. My 7D doesn't now the Tamron 1.4x is there when I use it with my EF 500/4.5L
    Could be a bug in Canon's firmware for the 1D-X. That's more likely than them deliberately enabling AF with a 3rd party TC. Maybe they'll fix the "bug" in the next firmware revision!
     
  3. It's because the 1.4x converter he's using lies to the camera body, which thinks his 800/5.6 + 1.4x is still a F5.6 lens. This trick has been used before with EOS bodies ...
    The real question is why it's not able to lie successfully to the 5D3 ...
     
  4. This comment by Arthur is weird:
    "ISO 400: Evaluative metering -1 stop: 1/640 sec. at f/9. Why -1 stop? The Canon system does not note the loss of one stop of light from the Kenko TC"

    Um. how can it not notice? Less light's coming through the lens, older metering systems just measured the light without using max aperture information from the lens! This just seems weird.
     
  5. It's possible that metering with the TC may be off. I remember that I used my Tamron 1.4x on my EF 500/4.5L with a previous camera (maybe EOS 20D or 40D) and it did need some exposure compensation dialed in. With the 7D it doesn't.
    I remember that the old 7e (the film camera!) need as much as 3 stops of EC when using some manual focus lenses. I think it depends a lot on the optics of the AE sensors.
    The reason why the TC needs EC could be the same reason that some manual focus lenses need exposure compensation, i.e. the position of the exit pupil of the lens shifts and that confuses the metering. If it happens with the Canon TCs, they can electronically communicate any compensation required to the body, so it's all transparent to the user. It's also possible that it's not sending some signal to the body that it normally gets from the lens which shifts the metering to compensate for the lens optics. Who knows what the lenses and camera bodies are talking about. I think there's more going on than just a report of an aperture number and driving the AF motor. The body is probably reading data out of a ROM in the lens.
    There's something odd going on because the correct focal length and aperture apparently get recorded in the EXIF data with the 1D X. It's almost as though the 1D-X reads that data and ignores it, while the 5D3 reads the data and says "Whoa, that's an f8 lens, I'm turning off the AF". Again, it could possibly be an unintended consequence of a firmware bug.
     
  6. It's my understanding that AF capabilities have relatively less to do with the amount of light streaming through the lens and much more to do with the differences in ray path angles. At wider apertures, there's a bigger difference in incident angles from opposite sides of the rear element to the focal plane; thus, the phase detectors can achieve a more accurate focus. It would be similar to comparing the accuracy of focus of two rangefinder cameras, one having a greater distance between the rangefinder windows. The manufacturers decide their phase detectors are sufficiently accurate to a certan cutoff aperture, and that's the aperture at which they disable the AF. But again, it's not as much about light. I can focus some of my cameras in light so dim that I'm not able to focus manually, even without a focus assist beam. (Admittedly it's a bit hit and miss.)
     
  7. Sarah's answer would seem to make some sense. An f8 or even f11 lens in bright light will have a much higher EV than an f2.8 lens in a dim room that can still AF.
     
  8. That's not the point. It's well known that illumination level isn't the reason for the f5.6 cutoff. Take that as read. Non-issue.
    The issue is why the 1D X and 5D MkIII are different since they use identical AF sensors. Their firmware is obviously different. The 5D MkIII acts as you would expect an EOS body to do, the 1D-X doesn't. Whether that is a feature or a bug probably depends on whether you're Canon or a user. If an EOS body with an f5.6 cutoff detects a TC has been mounted it cuts off AF. Since the 1D-X writes correct EXIF data to images shot with the TC (according to the report), it appears to know a TC has been attached, yet it doesn't cut off AF like it should. Canon would not do this deliberately for 3rd party TCs, yet retain the f5.6 limit for their own TCs. Either the 5D3 and 1D-X are looking at different data coming from the lens, or the 1D-X firmware isn't catching the TC like it should ("should" from Canon's viewpoint).
     
  9. The more interesting question is that if the AF functions successfully under these conditions, why does Canon disable it? A lot of buyers would like to keep f/8 autofocus. Perhaps the Canon 1.4x teleconverter, having different optics to the Kenko one, would not work with the focus points in the new body?
    Perhaps a 'sticky tape modification' can be done to the Canon TCs to keep autofocus at f/8 on the newer bodies.
    OTOH, I have seen that an older 1Ds body paired with a Sigma 180/3.5 lens and Sigma 2x teleconverter would not autofocus, even though the body is supposed to autofocus up to an effective aperture of f/8. This is with third party optics, of course, so from Canon's point of view all bets are off.
     
  10. Whether that is a feature or a bug probably depends on whether you're Canon or a user.​
    OT: I used to work in IT. As a QA I used to find many bugs and often the reply from the developers was: It's not a bug, it's a feature. As you can guess, we had a lot of quarrels about it. :)
    Happy shooting,
    Yakim.
    00an9Z-495049584.jpg
     
  11. bob:
    "The reason why the TC needs EC could be the same reason that some manual focus lenses need exposure compensation, i.e. the position of the exit pupil of the lens shifts and that confuses the metering."

    Ok, that makes some sense to me. Some :)
     
  12. Canon disable AF at f8 because it doesn't meet their specifications for speed and accuracy under their test conditions.
    It will work with some lenses under some conditions, as has been demonstrated numerous times with other lenses and other EOS bodies.
    Presumably Canon do not want nit picking AF complaints from those who would use AF at f8 and have it fail on them. As a manufacturer you really can't offer a feature that "sort of works some of the time with some lenses", even if you tell the user that.
    When I talked with Canon tech reps about the 5D MkIII they said they had the choice between (a) better AF at f5.6 and faster OR (b) AF to f8. One or the other, but not both. Choice (a) was chosen as it was likely to give the most benefit to most users. I'd guess there are a pretty small number of users trying to put a 1.4x TC on an 800/5.6. There are a few more trying to put a 2x on a 500/4 or 600/4, but tens of thousands of users using lenses faster than f5.6 including those using 2x TCs on lenses like the 300/2.8 and 400/2.8.
     
  13. Presumably Canon do not want nit picking AF complaints from those who would use AF at f8 and have it fail on them. As a manufacturer you really can't offer a feature that "sort of works some of the time with some lenses", even if you tell the user that.​
    But isn't that exactly what Nikon is doing?
    Happy shooting,
    Yakim.
     
  14. There are so many focus points it is strange Canon couldn't provide one measly focus point that copes with f/8 and have the other umpteen points use the design which gives faster general purpose use but only goes to f/5.6.
     
  15. Slightly off topic....
    On the finale of HBO's HARD KNOCKS they prominently featured a sports portrait photographer using a 1D X.
    Hard Knocks 2012 is a reality show in five parts about the Miami Dolphins NFL training camp that concluded in real time, just 2 days ago.
     
  16. Awesome. I hope all the nay-sayers who repeatedly insisted that there was no way for Canon's phase AF to work reliably at f/8 are reading this. Like anything, there is no "magic" line between f/5.6 and f/8 that makes phase AF inoperative. As the f-number goes up, the phase AF will get less and less reliable. However, I think this article is a clear example Canon withholding a basic feature because of the bizarre fear that someone shooting with $10,000+ in equipment (camera+lens+extender) will not comprehend that phase AF will not be perfect at f/8.
     
  17. As a manufacturer you really can't offer a feature that "sort of works some of the time with some lenses", even if you tell the user that.​
    Yeah sure. That is why they offer the annoying strobing flash on consumer cameras as an AF assist that fails nearly every time or numerous other features that require user no-how to be of benefit but may otherwise worsen a picture in the hands of the ignorant.
     
  18. Look at the price difference it is double, Canon 1D 2X will autofocus at F16. You got what you pay for.
     
  19. This makes sense based on my past experience with the Kenko Teleplus Pro DGX series. I own the Canon 2x II, the Kenko DGX 2X and the older DG version. My tests were with the Canon 70-200 f/4L USM Lens
    • The DG version would disable autofocus on my 7D (but not my 1D Mark II as you'd expect)
    • The DGX version would not disable autofocus on my 7D, would report the correct aperture to the camera and the EXIF data was correct (f/8) but required me to set EC to -2
    • I've come to believe that the DGX teleconverters do not disable autofocus but when used with some lenses designed to be used with teleconverters they require EC to be set to 1 or 2 stops depending on whether it's a 1.4x or 2x converter. It almost seems as though the camera double counts the teleconverter and over compensates in metering. This does not seem to be the case when used with my Canon 135 f/2 however.
    • When I used the DGX versions with lenses that are not designed to be used with teleconverters (My 85mm f/1.8 + Kenko 1.4x is really quite good) no EC is required to get correct metering.
    My tests this morning with my 5D Mark III also showed no problem with autofocusing at f/8. I was using a Sigma 70-200 f/2.8 plus a 1.4x and 2x Kenko teleconverters. The number of points accessible to me was less than the regular 61 but the camera was able to autofocus quite accurately. The speed was not as good as just the lens itself. I also tried my Canon 135 f/2 with the Kenko 1.4x and it functioned exactly as you'd expect.
     
  20. Brian, that's incredible. 10X for posting.
    Happy shooting,
    Yakim.
     
  21. The 5D3 will AF @ f/8 with the Kenko 1.4x. I just used it mounted on the 800 5.6 and it worked fine. Also works with
    updated firmware and it doesn't seem to overexpose as the 1DX does. About to go take it for a test drive on some
    wildlife. I'll post photos when I get back.
     
  22. Ok, so I'm back from my short test and here is what I found. I tested the 5D3 and 1DX side by side both with the 800mm f/5.6L IS and Kenko C-AF 1.4x Teleplus Pro 300. The resized photos are below. I set both cameras to ISO 1600, evaluative metering mode and Av. The 5D3 DID NOT overexpose the images with the 1.4x mounted as the 1DX does. However, in low light the 5D3 did not AF as well as the 1DX. This shouldn't be a disappointment given that the camera was not designed to even focus at f/8, but instead a bonus that the 1DX performs so well with f/8 AF. Now these aren't Nat Geo photos, just what I could get in a short trip down the road. Also, this is independent of the lens and t/c combo, but the AWB on the 1DX set for a cooler color temp than did the 5D3.
    00anb5-495501584.jpg
     
  23. The 5D3 had trouble focusing on the eye of the above photo (with both regular and spot AF) as the heron was in the shade. I had to put the sensor over the lighter streaked feathers on its back to gain focus. The 1DX focused quite well with the photo below. I used the spot AF on the heron's eye.
    00anbA-495503584.jpg
     
  24. And the 1DX with -1 EC
    00anbB-495505584.jpg
     
  25. In the shade, the exposure problems didn't seem like as big of a deal, but in harsh sunlight it was more apparent.
    00anbD-495507584.jpg
     
  26. And the 1DX...
    00anbE-495507684.jpg
     
  27. -1 exp. comp.
    00anbH-495507784.jpg
     
  28. Also, all photos above were shot in RAW, converted to DNG, imported to LR, and exported directly to Jpeg at 700p with no adjustments made. Also all were Av at f/8.
    So my conclusion, if anybody cares, is that the -1 exp. comp. required by the 1DX is not a big deal, unless you like to use auto ISO, in which there is no exp. comp. I know many people say auto ISO is useless, but I use it quite a bit as well as other wildlife shooters when we want complete control over f stop and shutter speed in changing light. The 5D3 wins in the exposure department and the 1DX wins in the AF department.
     
  29. Also found it interesting that the exif data on both cameras reported: f stop: f/8, and max aperture: 6. I don't know if this is odd or not, but surprised me that both cameras were fooled by max aperture but only one had exposure problems.
     
  30. We care, we definitely care... :)
    Happy shooting,
    Yakim.
     
  31. Also, let me clarify about auto ISO; there is no exp. comp. when used in M mode, which is most useful for ISO (it is available in Av and Tv). And one more thing, since the cameras weren't designed for f/8 AF, all focus points are available, unlike the 1D and 1DS series where only center point AF is available at f/8.
     

Share This Page