7D Mark II low light AF, an informal test VS. Fuji X-T1 and Olympus E-M10

[Aglet]

Just got my 7d2 recently and haven't had time to play with it.
So I just did some very informal quick AF testing outside. I compared it with my Fuji XT1 and my Olympus EM-10.
Lenses used were Canon EF 24-70mm f/4 L IS, Fuji XF 55-200mm f/3.5-4.8 OIS, and Panasonic 45-150mm f/4-5.6 respectively.
I used the Canon lens at 70mm, same with the Fuji, and the MFT system I set to about 55mm to provide similar angles of view.
All shots were hand held, Aperture priority mode, iso manually set to 12800, no EV comp and center AF points.

My back yard's fairly dark at night; ~ EV 0 to EV -1 where I was shooting.
exposures were:
1/8 @ f/4.4 @ 12800iso Oly EM10
1/3 to 1/2 @ f/4 @ 12800iso Fuji XT1
1/3 to 1/2 @ f/4 @ 12800iso Canon 7d2

What I found was the 7d2 struggled a little to AF in that light with that lens. 2 out of 3 test subjects it succeeded but it was the slowest camera of the 3 to focus in those conditions. The 3rd subject was too dark and low contrast for it to CDAF and it was completely useless in live-view. I think my old 5d2 had decent live view under such conditions.
I was able to effect a rough manual focus however.

The Fuji XT1 was able to provide a bright, if noisy, display of the scene in its EVF and it nailed 2 out of 3 test subjects very quickly and allowed me to MF the 3rd one but it was not easy or very successful.

The Oly also nailed the first 2 test subjects, almost as quickly as the Fuji, but its display was much dimmer and was completely unable to provide any assistance on the 3rd subject. It seemed to gain-up in a monochrome mode during AF of the first 2 subjects.

So pushing the low light AF limits with these 3 cameras tells me something I didn't expect.
Fuji's flagship, at 2/3 the price of the 7D2, was able to AF in dimmer light using its sensor-based AF system faster than Canon's latest and greatest AF system. Something worth considering if you shoot in dim venues.
The budget level Olympus EM-10 performed better than I thought it would under those conditions.

I already know the frustrating lack of speed these 2 mirrorless cameras have when used with long lenses even in good daylight but I am keen to see what the 7d2 will do with my old 100-400mm L lens in daylight and twilight conditions. That is what I bought it for and I hope it will perform as well or better than my original 7D under those conditions.

The backyard test subjects were, in order:
1. a galvanized pole reflecting a light source with a dark background in the distance
2. a dried thornapple seed pod with a dark background in the near distance, and the most difficult was
3. a mass of mountain ash twigs and leaves. Images of them below are reduced from the 7d2's ooc jpgs at 1/8 linear scale.

 

[sanj]

Thank you for posting the tests.

Am wondering why the images are not sharp. Thx.

 

[Mt Spokane Photography]

You might compare using the 7D MK II sensor based system(Live AF), that might provide a better comparison when comparing with a different cameras sensor based AF system.

 

[Aglet]

You might compare using the 7D MK II sensor based system(Live AF), that might provide a better comparison when comparing with a different cameras sensor based AF system.

I actually wanted to compare the latest PDAF system of the 7D2 to the hybrid AF system of the Fuji and the purely CDAF system of the little Olympus.
FWIW I did go back out last night to try live-view AF on the 7D2 and it failed to gain-up enough to even see the 3rd subject.. nothing, nada, black screen. As as bad as the Oly performed in that situation. I did not bother with trying it on the slightly brighter scenes. It may have worked on those since there was greater contrast but that wasn't what I had intended to test.
I remember that one of the things I really like about my old 5D2 was that in very low light scenes like this, live-view would gain-up very well to at least provide a noisy but usable view where one could effect manual focus. The Fuji's EVF performed very well in this situation but still made it difficult to judge when the leaves were in good focus as the MF ring on the lens is frustratingly slow in response. Fuji users are hoping they'll add an acceleration curve function to the MF ring to make MF easier and now I can see why.

 

[Aglet]

Thank you for posting the tests.

Am wondering why the images are not sharp. Thx.

I'd venture they're not sharp because I was well beyond the lowest shutter speed/focal length where the IS is able to deliver. Combine that with medium focal length, relatively shallow DoF, any bit of motion is going to cause some blurring. Add in the level of way late night fatigue I had there was no way I was going to be as steady as I could have been either.
None of the images were sharp, motion blur seems to have gotten most of them far more than the in-camera NR for loss of detail.

The first 2 shots with the 7d2 were at 1/4 second and 70mm - that's between 4 & 5 stops slower than would be recommended as a slowest shutter speed for that focal length to have a reasonable chance of a clear image and likely a stop or 2 beyond what most IS systems can deliver.
The 3rd shot was at 0.4 seconds so nearly 1 stop slower yet. I'm not sure if there was also some breeze-induced motion on the leaves there as some are much more blurred than others. It was just hard to see and judge focus on top of that.

I've now scaled the Fuji and Oly shots to similar size for comparison.. (hmmm.. musta been a late nite, couldn't even keep my landscape=portrait orientation consistent!)

The reason I even started this test was I noticed the 7d2 AF acting unexpectedly in some low light indoors situations. I was merely playing with the lens to see what its close-focus abilities were like. I was aiming at a bright stripey textured throw rug with lots of color and contrast and well within the min focus distance when I found the 7d2 refused to AF. It would just hunt, with in-focus quickly passing thru as it racked each way. None of my ML cameras had any issues with this and I don't recall any of my other SLRs struggling with such a focus target. I'll test more if I have time in the next few days as I was puzzled and not impressed by that example.


I can't repeat these exact tests as we just had a dump of snow so the lighting and targets are now quite different.

 

[Aglet]

Here's the shots from the Olympus E-M10.
All at 53mm, 1/8s, f/4.4, 12800iso, handheld and set to use the IS of the lens rather than the in-body sensor-based IS. As you can see, the 3rd shot of the leaves was hopelessly OOF while sharpness on the other 2 is better.

 

[Aglet]

Here's the shots from the Fuji X-T1

All at 70.5mm, f/4, 12800iso (Fuji's ISO may be about 1 stop optimistic)
5/13s shutter for the pole
1/3s shutter for the thornapple
1/2s shutter for the leaves

 

[Woody]

2 questions:

i) Was that done with centre AF point on 7D2?

ii) Did you change the EVF setting on the EM10 so that the VF will gain in brightness? I understand the EM10 EVF can be set to simulate brightness of subjects to the human eye.

 

[ajfotofilmagem]

All at 53mm, 1/8s, f/4.4, 12800iso, handheld and set to use the IS of the lens rather than the in-body sensor-based IS.

I appreciate your initiative. ??? But without a tripod, your test is just a random result and is not "repeatable" as tests they should be. Blur when holding the camera in hand prevents the AF system to work properly. :-\

 

[Aglet]

i) Was that done with centre AF point on 7D2?

yes (last line, first paragraph, opening post )

ii) Did you change the EVF setting on the EM10 so that the VF will gain in brightness? I understand the EM10 EVF can be set to simulate brightness of subjects to the human eye.

I have EVF auto-luminance set to ON, relative exposure simulation is also active.
I think i just wasn't providing enough signal to work with at the framerate the EVF operates at.

 

[Aglet]

All at 53mm, 1/8s, f/4.4, 12800iso, handheld and set to use the IS of the lens rather than the in-body sensor-based IS.

I appreciate your initiative. ??? But without a tripod, your test is just a random result and is not "repeatable" as tests they should be. Blur when holding the camera in hand prevents the AF system to work properly. :-\

I understand your viewpoint here but the point of this informal test is to test the ability of these cameras' AF systems to perform in difficult conditions. Using a tripod should allow a more consistent test from shot to shot but handheld we get an idea of combined performance of the AF system and the optical stabilization systems.
This is not necessarily an unrealistic approach either. There are plenty of times that I'll grab a camera to take a shot of something in very dim lighting and do not have time to set up a tripod. using AF illumination is also not always an option, it may disturb the subject or give away your presence when trying to do a candid.

The stabilizers in each of the lenses should be adequate to allow AF operations handheld, that's one of their benefits.
The long exposure times required to make an image is a different matter, being mostly beyond currently available optical stabilization abilities. Tho FWIW, the Fuji's done a decent job on the 2nd subject.

 

[tapanit]

without a tripod, your test is just a random result and is not "repeatable" as tests they should be.

That is true enough, but:

Blur when holding the camera in hand prevents the AF system to work properly. :-\

If that is true, the camera is all but useless. Many, indeed probably most people will use it hand-held most of the time, and AF definitely should work properly then, even if it's harder to test.

 

[tayassu]

A little dirt to the nearly perfect image of the 7DII...
Hopefully, AF issues don't get in the way of this great body!!!

But thanks for sharing!