Canon 5D mark IV + EF 35 mm F/1.4L II USM autofocus problems

grinbo

I'm New Here
Apr 26, 2019
21
2
I have a question about autofocus in 5DM4. I mean not LiveView focusing. I have problem with refocusing during session of shots. In LifeView the lens focus perfectly. My test step: I put the camera on tripod -> focusin on really contrast image (test chart by Focal) using central focusing dot. Distance between camera and target - 50. For my lens it it 3550 = 1750 mm. For shot use remote shutter release. And if i put the autofocus button, some time lens refocusing. And not always return to it initial place -> result: back or front focus. 2 of 10 attempts failed. In LifeView mode lens focusing perfectly without any refocusing.
I've tried to colibrate lens using Reikan Focal software. AF auto colibration works ok. But result for focus consistency - 2 of 10 shots failed.
How can i determine where is a problem? Camera of lens?
In real life shots i think the problem is worse. May be it's my fault...
 

Viggo

EOS 5D SR
Dec 13, 2010
4,228
792
What I did, as the semi-automatic test also does, is defocus the lens for each shot. If it tries to refocus when it’s in focus first, it might not try or work reliable enough.

If that doesn’t do any difference I would take 20 shots and if they are different, or more than 2 of them are not the same, I would send the lens and camera to Canon.

I have absolute sympathy for your frustration, and these sorts of issues are one of the big reasons I went for mirrorless. It’s such a relief ...
 
  • Like
Reactions: caMARYnon

grinbo

I'm New Here
Apr 26, 2019
21
2
What I did, as the semi-automatic test also does, is defocus the lens for each shot. If it tries to refocus when it’s in focus first, it might not try or work reliable enough.

If that doesn’t do any difference I would take 20 shots and if they are different, or more than 2 of them are not the same, I would send the lens and camera to Canon.

I have absolute sympathy for your frustration, and these sorts of issues are one of the big reasons I went for mirrorless. It’s such a relief ...
I'm also think to go to mirrorless. :) For testing i'm using Reikan Focal software. It defocus lens for eacy shot. I can configure it. And my result is 2 of 10 - bad focusing. A attach results to this message.
I've read a lot articles about autofocus colibration. In most of them authors told, that i must use some special lights for testing (very close to daylight) and using recommended distance. But it not work perfectly in real life. I take pictures not only on the street, but also in rooms with a different light. And at different distances ...
 

Attachments

Viggo

EOS 5D SR
Dec 13, 2010
4,228
792
I'm also think to go to mirrorless. :) For testing i'm using Reikan Focal software. It defocus lens for eacy shot. I can configure it. And my result is 2 of 10 - bad focusing. A attach results to this message.
I've read a lot articles about autofocus colibration. In most of them authors told, that i must use some special lights for testing (very close to daylight) and using recommended distance. But it not work perfectly in real life. I take pictures not only on the street, but also in rooms with a different light. And at different distances ...
I’ve used FoCal for years, actually since it first came out, and it was great up until one update where from that and on it gave consistent front focus for me.

I’ve used all kinds of light, and if I used sunlight or my 500W workers lamp it gave the same exact results, so the light, even flickering lights, doesn’t matter as much as one would assume.

I also test IRL with my results and even though it might not be perfect, excluding Sigma, it has been very consistent, especially with the 35 L II.

I would send the combo to Canon.
 
  • Like
Reactions: cpsico and mikekeck

Maximilian

The dark side - I've been there
Nov 7, 2013
2,650
462
Germany
Hi grinbo!

I don't use Reikan Focal myself but from all I've read about the test procedures is that you really need three things for good (and reproducible) results:
  1. A really solid ground (no parquet, carpet etc.)
  2. A really solid tripod
  3. A lot of bright and constant bright light (indoors eg. 2 flood lamps)
You mentioned some of them above, I know.
But I wasn't sure if you understood that you do not reproduce the environment you normally shoot at, but you build up a test stand for calibration of your AFMA.
After finding the right AFMA correction value of your setup the AF results will (should) become better but not 100% perfect.
Especially if you shoot in dim light conditions where the AF always struggles

And if you cannot reproduce consistent results in this test stand conditions I tend to Viggos conclusion to send the combo to Canon.

I hope I could help a little bit.
 
  • Like
Reactions: Viggo

Viggo

EOS 5D SR
Dec 13, 2010
4,228
792
I flipped through your report, and did I read correctly that you had 1/25s shutter speed?

That can’t be 10-11 EV light at f1.4. Too low light will for sure produce unstable results.

Another thing is that the Quality of Focus is very high. Even though the graf seems to show huge variations in results from shot to shot, looking at the lowest and highest value they’re not that far apart.
 

grinbo

I'm New Here
Apr 26, 2019
21
2
Hi grinbo!

I don't use Reikan Focal myself but from all I've read about the test procedures is that you really need three things for good (and reproducible) results:
  1. A really solid ground (no parquet, carpet etc.)
  2. A really solid tripod
  3. A lot of bright and constant bright light (indoors eg. 2 flood lamps)
You mentioned some of them above, I know.
But I wasn't sure if you understood that you do not reproduce the environment you normally shoot at, but you build up a test stand for calibration of your AFMA.
After finding the right AFMA correction value of your setup the AF results will (should) become better but not 100% perfect.
Especially if you shoot in dim light conditions where the AF always struggles

And if you cannot reproduce consistent results in this test stand conditions I tend to Viggos conclusion to send the combo to Canon.

I hope I could help a little bit.
Hello. I have solid tripod, but on carpet. I don't think that it helps me to improve my result, but i'll try... It is my first lens with f/1.4.I've used max 2.8 before it. And haven't big problems like this.
 

grinbo

I'm New Here
Apr 26, 2019
21
2
I flipped through your report, and did I read correctly that you had 1/25s shutter speed?

That can’t be 10-11 EV light at f1.4. Too low light will for sure produce unstable results.

Another thing is that the Quality of Focus is very high. Even though the graf seems to show huge variations in results from shot to shot, looking at the lowest and highest value they’re not that far apart.
I'm using auto settings for shutter speed (AV) as Focal recommend. I'm not change any settings in camera. I've tested it at evening using simple room light on the ceiling. I will next test it using daylight from window :) Thank you for your review of my results.
 

Maximilian

The dark side - I've been there
Nov 7, 2013
2,650
462
Germany
Hello. I have solid tripod, but on carpet. ...
I've read about people going to the cellar to put their tripod directly on concrete. But of course in the cellar you have to take care of light even more.
I really hope it helps.
 
  • Like
Reactions: grinbo

Viggo

EOS 5D SR
Dec 13, 2010
4,228
792
A 500W lamp at 30 cm gives around 10-10,5 ev light which is the number you’re looking for. It’s also good to have the light evenly spread.

And that the target is matte and not glossy. I also like to have a black matte background to not create a reflection.
 
  • Like
Reactions: grinbo

Graphic.Artifacts

EOS 7D MK II
Aug 1, 2017
465
268
The purpose of AFMA is to correct for a mild persistent front or back focusing lens/body combination. In other words, the lens always front or back focuses by the exact same amount due to a calibration error between the AF system in the body and the lens. It will not help with a lens that has a focus precision problem which is a lens that either/both front and back focuses by varying amounts.

If you AFMA a lens with a precision issue some of your photos may get better but some will also be worse. AFMA has no ability to correct for AF precision issues which is the problem you seem to be describing.

If you are confident that you are operating the AF system properly and you are not getting precise focusing, I'd send it to Canon as the other posters suggested.

Despite what a lot of people here and elsewhere will tell you, the chances of you significantly improving the performance of a lens without a clearly identifiable and persistent front/back focusing problem with AFMA is slim to none.

When you put an AFMA value in your 5D4 all you are doing is telling the camera to intentionally front or back focus on every shot with that lens. If you aren't sure that's needed why would you tell the camera to do that? IMO one of the most mis-used features of Canon DSLR's.

The AF module in Canon's DSLR's is just inherently less precise than the PDAF used in live view and the EOS R's.
 
Last edited:

cpsico

EOS RP
Mar 27, 2011
356
10
I have used afma before as temporary patch until I can send it in to canon for calibration. It's always better, esp if the lense is under warranty.
 

Viggo

EOS 5D SR
Dec 13, 2010
4,228
792
I have used afma before as temporary patch until I can send it in to canon for calibration. It's always better, esp if the lense is under warranty.
The issue with that is that they move the AF points further or closer to compensate, so when you keep the lens and switch the body it’s off again. But it is accurate
 

cpsico

EOS RP
Mar 27, 2011
356
10
The issue with that is that they move the AF points further or closer to compensate, so when you keep the lens and switch the body it’s off again. But it is accurate
This is probably the biggest reason mirrorless will kill dslr cameras. None of these focus calibration issues on your expensive lenses
 
  • Like
Reactions: caMARYnon and Viggo

Viggo

EOS 5D SR
Dec 13, 2010
4,228
792
This is probably the biggest reason mirrorless will kill dslr cameras. None of these focus calibration issues on your expensive lenses
Yeah, that was a big reason for me at least, I hate to pay for sharpness you don’t really get or not all the time.

The R and RF50 combo is incredible when it comes to hitrate.
 
  • Like
Reactions: cpsico

AlanF

Canon 5DSR II
Aug 16, 2012
5,806
3,296
This is probably the biggest reason mirrorless will kill dslr cameras. None of these focus calibration issues on your expensive lenses
A mirrorless that uses contrast detect does not require AFMA. But, if only phase detect is used, like for the EOS R, then some AFMA may be needed because of errors in lens centering, errors in the angle of the planes of the mount and sensor etc. Errors in alignment of an adaptor could skew tthe AF. The range of AFMA required should be much less than that for a mirror system, but AFMA may still be required to optimise performance. Some Olympus and Sony bodies do have AFMA.
 
  • Like
Reactions: Michael Clark

Mt Spokane Photography

I post too Much on Here!!
Mar 25, 2011
15,547
770
I'm using auto settings for shutter speed (AV) as Focal recommend. I'm not change any settings in camera. I've tested it at evening using simple room light on the ceiling. I will next test it using daylight from window :) Thank you for your review of my results.
Even daylight from a window is probably not bright enough, you need really bright light. If you are in a apartment building, its even more important, because even concrete floors vibrate and move. You want the shutter speed to be as high as possible, at f/1.4 Iso 100 and bright EV 13 light, shutter speed should be about 1/4000 sec. There is some leeway, but with a very fast lens and ISO 100, get at least enough light for over 1/500 sec shutter.
 

cpsico

EOS RP
Mar 27, 2011
356
10
A mirrorless that uses contrast detect does not require AFMA. But, if only phase detect is used, like for the EOS R, then some AFMA may be needed because of errors in lens centering, errors in the angle of the planes of the mount and sensor etc. Errors in alignment of an adaptor could skew tthe AF. The range of AFMA required should be much less than that for a mirror system, but AFMA may still be required to optimise performance. Some Olympus and Sony bodies do have AFMA.
Good to know, thank you. I have used the contrast based live view on my 6D and found it to be really accurate and useful in some situations, backlit not being one of them, but still nice for portraits
 

JuanMa

Canon New F1
May 10, 2018
65
53
Spain
I also get inconsistent results with Reikan (probably my fault). What I usually do is, first, take some shots in live view mode using a Spyder lenscal as target and then proceed to microadjust the lens till I get the same results than those taken in live view mode. That works perfectly for me. It doesn't take more than 10 min. per lens.