AF problems with 70D and fast lenses

[Lightmaster]

http://www.traumflieger.de/reports/Kamera-Tests/Canon-EOS-DSLR/EOS-70D/EOS-70D-Autofokus-Problem-Stellungnahme::617.html?XTCsid=a30801259ebc5a51cdc541a02e838980

not the best source, i know

but i have read a lot about it on german forums.
this video maybe is interesting for some 70D users.

the 70D in the video clearly has a AF problem using the center AF point with the 24mm f1.4.
i have heard about similiar problems with f1.8 and even f2.8 lenses and the center AF point.

 

[Famateur]

Thanks for the link. Anyone care to translate the gist of the article/video into English?

I have a 70D and EF-S 17-55 F2.8 IS, and I've noticed that focus is hit or miss via the viewfinder -- mostly miss. I've invested in FoCal and run through the calibration process, and each time it produces a different AFMA value for a given focal length and distance (e.g. running the calibration process several times in a row without changing any of the setup). The suggested AFMA values are usually anywhere from -1 to +2 or +3.

Calibration Info

1. Tripod on concrete basement floor at night.
2. Constant lights providing ~12ev.
3. Calibrations at distances of both 25x and 50x focal length
4. Calibrations for both 17MM and 55MM focal length

On one occasion, FoCal did not complete the calibration saying that the results were not consistent enough to generate an AFMA value, although the two or three calibrations I did after that (with identical setup) worked with "Good" results.

Unfortunately, I don't have another lens to test with right now, so I'm left wondering if it's the body or the lens. Live view is usually correctly focused, which is what pointed me to AFMA to begin with, but even after calibration, focusing through the viewfinder is unreliable enough that I'm mostly using Live View now so I don't miss the shots.

Is this likely a body or lens issue (or user error )? It's just frustrating when I focus on an eye (at shutter speeds well beyond 1/FL) and get a nice sharp eyebrow but blurry eye or even nothing sharp at all. Maybe I should test individual focus points?

Anyway, this is something I'm trying to get to the bottom of -- when I have the time. Any tips or info are welcome.

Cheers...

PS -- One other thought: while it's probably far less likely, is it possible that there's a slight misalignment of either the AF sensor or the LCD overlay in the viewfinder such that the focus point I see isn't really representing the location that the AF sensor is "seeing"? For example, what if the AF sensor really is seeing the eyebrow when the focus point in the viewfinder is over the iris? Just thinking out loud here...I should probably set aside an afternoon, channel my inner Neuroanatomist and do this in as structured and scientific a manner as possible...and maybe borrow another lens while I'm at it.

 

[Lightmaster]

in short, using the center AF point on the 70D with some(!) lenses and wide open aperture will result in out of focus images. no amount of AFMA adjustment will help.

choose a different AF point and the lens will focus fine.

seems to be a strange issue because some people report that the same kind of lens works fine on their 70D, while other complain that they can´t get a single sharp photo (at wide open aperture).

in the video above only the 24mm showed this behavior.
but i read reports about the same behavior with other fast glass.
the guy in the video says he noticed this with the 100mm L macro earlier on a 70D ... but not this time.

 

[Famateur]

Thanks for the summary!

I'll add this info to the context of my ongoing testing...

Thanks again!

 

[retina]

Could an issue like this be solved with new firmware?

 

[Famateur]

Doing a little searching of the interwebs, I found this: http://photo.stackexchange.com/questions/43791/why-isnt-my-canon-70d-autofocus-accurate-in-manual-zone-af-mode-with-a-50mm-f-1

Comments on that post by jrista (I'd assume the same individual as CR's jrista) were very helpful. It might not solve my issue, but it definitely helps frame my perspective as I approach my testing.

The explanation of the 7/70D AF sensor (linked below) was also helpful. In fact, based on the explanation that the sensor lines extend beyond the boundary of the viewfinder's focus square, I might not be too far off on my question about discrepancies between what the viewfinder sees and what the AF sensor sees.

http://photo.stackexchange.com/questions/41174/if-the-focal-plane-is-curved-should-the-outer-af-points-work-correctly-or-front/41179#41179

Anyway, interesting stuff. How I'd love to have a 5DIII AF system, but alas, the funds are lacking...

 

[MichaelHodges]

This is a bummer, but I'm not surprised. At all.

 

[East Wind Photography]

Doing a little searching of the interwebs, I found this: http://photo.stackexchange.com/questions/43791/why-isnt-my-canon-70d-autofocus-accurate-in-manual-zone-af-mode-with-a-50mm-f-1

Comments on that post by jrista (I'd assume the same individual as CR's jrista) were very helpful. It might not solve my issue, but it definitely helps frame my perspective as I approach my testing.

The explanation of the 7/70D AF sensor (linked below) was also helpful. In fact, based on the explanation that the sensor lines extend beyond the boundary of the viewfinder's focus square, I might not be too far off on my question about discrepancies between what the viewfinder sees and what the AF sensor sees.

http://photo.stackexchange.com/questions/41174/if-the-focal-plane-is-curved-should-the-outer-af-points-work-correctly-or-front/41179#41179

Anyway, interesting stuff. How I'd love to have a 5DIII AF system, but alas, the funds are lacking...

I have similar issues with the 5d3. I discovered some things

1. The 24mm and in fact many lenses have quite a different afma depending on focus distance.
2. Focal is not very reliable. I have never obtained consistent results using my 5diii or 7d with 10 different lens combinations.
3. Fast lenses have a very thin dof so any error is more apparent. You can try the lenses at f4 and see if your curves stabilize.

I ended up having to manually verify focal with a Spyder lenscal and ended up just not using focal as i got better results doing it manually.

 

[Lightmaster]

again this is not a front/backfocus issue that can be easily solved with AFMA.

it´s also not an AF tracking issue, or something like that.
when your combo is affected by this problem each and every image will be out of focus at wide open aperture. as you see they use a tripod and a static object (tree).

when you change from center AF to a different (not center AF) point the lenses behave just fine.
again.. that makes AFMA pretty useless.

the lenses are also not faulty per se.
as shown the same lens works fine on 60D´s or 7D´s (center AF point or not doesn´t play a role).

 

[ishdakuteb]

i would challenge him to wait for another > 2 hours or even might be 3, try it with the same angle/position.... then post video here for me to see the result... just saying...

 

[MichaelHodges]

BTW, I've heard many complaints about the 24 1.4 well before the 70d came to fruition.

 

[Famateur]

again this is not a front/backfocus issue that can be easily solved with AFMA.

it´s also not an AF tracking issue, or something like that.
when your combo is affected by this problem each and every image will be out of focus at wide open aperture. as you see they use a tripod and a static object (tree).

when you change from center AF to a different (not center AF) point the lenses behave just fine.
again.. that makes AFMA pretty useless.

the lenses are also not faulty per se.
as shown the same lens works fine on 60D´s or 7D´s (center AF point or not doesn´t play a role).

Understood. I apologize if my particular (separate) issue hijacked the thread. I wrote about it before I knew the contents of the video. Thanks for the clarification!

 

[Famateur]

Doing a little searching of the interwebs, I found this: http://photo.stackexchange.com/questions/43791/why-isnt-my-canon-70d-autofocus-accurate-in-manual-zone-af-mode-with-a-50mm-f-1

Comments on that post by jrista (I'd assume the same individual as CR's jrista) were very helpful. It might not solve my issue, but it definitely helps frame my perspective as I approach my testing.

The explanation of the 7/70D AF sensor (linked below) was also helpful. In fact, based on the explanation that the sensor lines extend beyond the boundary of the viewfinder's focus square, I might not be too far off on my question about discrepancies between what the viewfinder sees and what the AF sensor sees.

http://photo.stackexchange.com/questions/41174/if-the-focal-plane-is-curved-should-the-outer-af-points-work-correctly-or-front/41179#41179

Anyway, interesting stuff. How I'd love to have a 5DIII AF system, but alas, the funds are lacking...

I have similar issues with the 5d3.

This actually cheers me up a bit.

1. The 24mm and in fact many lenses have quite a different afma depending on focus distance.
2. Focal is not very reliable. I have never obtained consistent results using my 5diii or 7d with 10 different lens combinations.
3. Fast lenses have a very thin dof so any error is more apparent. You can try the lenses at f4 and see if your curves stabilize.

I ended up having to manually verify focal with a Spyder lenscal and ended up just not using focal as i got better results doing it manually.

Interesting about different AFMA at different focus distance. I'll test mine to see if it's similar. After shelling out for FoCal, I'm a bit disappointed to hear your results. I'm going to give it another go with as meticulous and repeatable a setup as I can before I give up on it (and maybe with a couple of borrowed lenses). Agreed on the DOF challenges. I first thought that my issues were from moving forward/back too much between focus lock and shutter release, but I didn't experience any issues via Live View.

Thanks for the input! I'll be sure to report back and/or ask further questions in a new thread (didn't mean to hijack this one).

 

[Lightmaster]

i would challenge him to wait for another > 2 hours or even might be 3, try it with the same angle/position.... then post video here for me to see the result... just saying...

BTW, I've heard many complaints about the 24 1.4 well before the 70d came to fruition.

did you guys actually read the thread before posting?

when you change from center AF to a different (not center AF) point the lenses behave just fine.

so what will waiting a few hours change on the center AF point?
why does the 24mm (and other fast lenses) works fine with AF points other then the center AF point?

 

[ishdakuteb]

so what will waiting a few hours change on the center AF point?

that is something relates to light (it is also your user name). i stopped watching video at 10:45 after clearly identified:

1. focus object,
2. direction of light,
3. position where he stood and camera focus angle,
4. and some other, but the top 3 are much much important...

well, this is my own guess according to my understanding. but i would love to see second video demonstrated like i have described if i can...

 

[ashmadux]

@ OP

You should refer to the review here of the 24 1.4. He does mention focus issues, but even i didn't really notice that line until after I tried it myself. I rented this last summer, trying to somewhat replicate my great experience with the EOS - M's 22/f2 (love it!) on a full sized body.

Spoiler: it didnt go well. I had problems focusing on pretty much anything in the most amateur of shooting situations. And I mean anything- people, places, and things. It was all over the place. OOF 99% of the time- i couldnt even get a bridge in focus- not kidding.

I can only talk about my experience, but im guessing that this is pretty well known. Maybe.

 

[MYB]

I'm not sure but i have this problem too imo. I have 70D with sigma 17-70mm f/2.8-4. There is a focus problem at wide angles (17-20mm). Maybe because of the lens but i have only one, couldn't try with another. AFMA didn't solve it. I'm using in aiservo mode with all 19 points, best i can for now.

 

[drjlo]

I have similar issues with the 5d3. I discovered some things

1. The 24mm and in fact many lenses have quite a different afma depending on focus distance.
2. Focal is not very reliable. I have never obtained consistent results using my 5diii or 7d with 10 different lens combinations.
3. Fast lenses have a very thin dof so any error is more apparent. You can try the lenses at f4 and see if your curves stabilize.

I ended up having to manually verify focal with a Spyder lenscal and ended up just not using focal as i got better results doing it manually.

I have similar issues with 5D III with its "Advanced 61 AF points", something I'm sure is prevalent with all phase-detect/mirrored bodies out there. There is good reason live view with magnification is recommended for critical focus. The problem is current DSLR's are slow as molasses to AF in live view.

I do think the future is mirrorless bodies with advanced AF, not having to go through the mirrorbox mechanism, such as the upcoming Sony A6000. Having tried the Sony A7R, I do look forward to the day the A6000 AF system is implemented on that A7R sensor.

 

[East Wind Photography]

I have similar issues with the 5d3. I discovered some things

1. The 24mm and in fact many lenses have quite a different afma depending on focus distance.
2. Focal is not very reliable. I have never obtained consistent results using my 5diii or 7d with 10 different lens combinations.
3. Fast lenses have a very thin dof so any error is more apparent. You can try the lenses at f4 and see if your curves stabilize.

I ended up having to manually verify focal with a Spyder lenscal and ended up just not using focal as i got better results doing it manually.

I have similar issues with 5D III with its "Advanced 61 AF points", something I'm sure is prevalent with all phase-detect/mirrored bodies out there. There is good reason live view with magnification is recommended for critical focus. The problem is current DSLR's are slow as molasses to AF in live view.

I do think the future is mirrorless bodies with advanced AF, not having to go through the mirrorbox mechanism, such as the upcoming Sony A6000. Having tried the Sony A7R, I do look forward to the day the A6000 AF system is implemented on that A7R sensor.

I agree. Future of af is to do it in the sensor where the image is obtained. Need to speed up af times, faster than phase detect, and do it with an accurate viewfinder for object tracking. 7dii may try to address these issues but i doubt the af speed will improve much....which is why it will first be marketed for video use.

As far as fast lenses go its best to afma them at the focus distance you expect to use the lenses the most. Outside of that, use manual or live view focus, or use a higher f stop to increase dof. Id like to see afma tweaked to allow for afma at distance and minimum focus distance then extrapolate between the two. Similar to what they do now with zoom lenses at wide and tele. That should help considerably.

 

[Lightmaster]

so what will waiting a few hours change on the center AF point?

that is something relates to light (it is also your user name). i stopped watching video at 10:45 after clearly identified:

1. focus object,
2. direction of light,
3. position where he stood and camera focus angle,
4. and some other, but the top 3 are much much important...

well, this is my own guess according to my understanding. but i would love to see second video demonstrated like i have described if i can...

sorry but you missed the point entirely.

as i wrote a few times: the camera focuses fine with any other focus point then the center AF point.... under the SAME light situation.

there goes your theory...

IF your theory would hold it´s water then all AF points should be affected... you agree?