I already read that the autofocus of non-STM lenses is rather slow on the M models, but I noticed that my Canon EF 50mm/1.4 is noticeable slower in autofocus than my EF 28mm/1.8. Also a EF 100mm/2.8 is much faster - all of them have USM and are very fast on my 5Dmkii.

As the 5Dmkii uses phase detection autofocus and the M3 has contrast detection AF, is the larger aperture here of disadvantage? As a sharp edge can not be detected so easily with the smaller depth of field?

  • 1
    Are you concerned with the autofocus speed or the sharpness of the image? I don't follow how the last part of your question is relevant.
    – dpollitt
    Jan 7 '18 at 20:57
  • I thought about that the autofocus is unable to detect sharp edges, as the field of depth is less with the 1.4 aperture. Usually I think the autofocus works better the larger the aperture, ie it usually works not so good on lenses with higher aperture numbers (for example a 2x teleconverter + 5.6 lens). But maybe not on the autofocus of the M3?
    – reox
    Jan 7 '18 at 21:09
  • To autofocus, Canon cameras use the widest aperture available then stop down to take the image as necessary.
    – dpollitt
    Jan 7 '18 at 21:10
  • yes. so is this of disadvantage on a mirrorless camera with only CDAF? edit: the M3 should also have a phase detection AF.
    – reox
    Jan 7 '18 at 21:17
  • Is the EF 50mm f/1.4 also much slower when using CDAF (Live View) with your 5D Mark II?
    – Michael C
    Jan 8 '18 at 15:50

The EF 28mm f/1.8 USM and the EF 100mm f/2.8 USM Macro both have ring type USM, as do the vast majority of Canon's lenses designated as USM.

The EF 50mm f/1.4 USM has a micro-motor type of USM that is pretty much unique to this one lens among Canon's current lens lineup.

In my experience, the EF 50mm f/1.4 is a bit slower than all of my other USM lenses when used on my Canon DSLRs. I've not shot with that lens mounted on any of the mirrorless EOS M models.

As the 5Dmkii uses phase detection autofocus and the M3 has contrast detection AF, is the larger aperture here of disadvantage? As a sharp edge can not be detected so easily with the smaller depth of field?

If the lens is way out of focus when CDAF begins it can take a little longer for the camera to decide which way it needs to go to increase contrast. Does this also happen when using the CDAF (Live View) on your 5D Mark II?

With the smaller form factor (and thus smaller battery), I wouldn't be surprised if the EOS M series provides less power to move the focusing elements of any EF lenses than the DSLR bodies do. It's fairly well known that the EOS 1-series cameras, with their much larger batteries, move AF lenses faster than other EOS DSLRs. The fact that ring-type USM probably takes less power to move a focusing element with the same mass than the micro-motor USM of the 50/1.4 does may be the greatest part of what you are noticing.

  • I tested both optics on my 5dmkii in live view. Both are equally bad there. They both take the same time to focus (I did not use a stopwatch...) and the focusing pattern is the same. They go from corse, to middle to fine steps. With the focus peaking of magic lantern, you can see this quite nicely. The AF pattern on the M3 is quite different to the one on the 5Dmkii. On the M3, the focus jumps back and forth in the beginning, this seems not to be the case on the 5d. In my opinion, the jumping back and forth with the focus is much quicker on the 28mm
    – reox
    Jan 10 '18 at 17:00

It's not up to the aperture, it's up to the method how the lens focuses.

The most of older (and not so old) (D)SLR lenses use unit-focusing method, they move the whole optic group to focus, while modern mirrorless lenses focus by moving only one lens, or just a small portion of the lens optics. Moving only one (or a few) lens(es) is much faster method than moving around the whole optical group of a lens, so it is expected that the unit focusing method would be slower - and from practical experience, I can say it most definitely is.

The vast majority of mirrorless lenses use the second method (moving only one lens) to achieve fast focusing speed, the only exceptions I can think of are the first two microFourThirds primes - Panasonic G 20/1.7 and Olympus M.Zuiko 17/2.8, both use a classic DSLR-style unit-focusing method. I've had 20/1.7, it was focusing slower than any other lens in the system, especially when compared to 25/1.4 or 45/1.8 lenses, and all those lenses have a large aperture.

Shallower depth of field should be easier for CDAF - there's more difference between what's sharp, and what isn't. On the other hand, any loss of contrast and sharpness many older optical designs have is a disadvantage.

It is not unexpected that EF50/1.4 DSLR lens focuses slowly - that's as a classical SLR-style lens as you can get, it is designed in 1993, in the film era while autofocus was still relatively new technology, in the time when nobody was pixel-peeping, and great sharpness wasn't expected at wide apertures.


Live View CDAF is significantly affected by the lens' max aperture (or current aperture if using exposure preview). This is because the DOF of the image on the sensor is directly affected by it.

This doesn't happen w/ PDAF because the DOF at the PDAF sensor is not affected by the lens' aperture setting. (I suppose it could be affected if the lens' max aperture were VERY small).

Additionally, the images used for CDAF are much larger than those used for PDAF which inherently reduces their "sharpness/DOF". But it also makes it possible for CDAF to focus on much smaller details (in general).

However, I would be surprised if the difference between f/1.4 and f/1.8 is enough to be of significant impact.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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