7D Mark II - Focus Discussions

Lens Drive When AF impossible
- Continue focus search ON
- Stop focus search OFF

I'm just back from the Everglades. I used the 7D MkII and original 100-400 extensively. Here is a link to keeper my shots: https://www.flickr.com …r/sets/72157650​467696926/.

Nearly all of the shots were in Aperture priority and AIServo mode. For static birds I used one point to focus, for birds in flight I used the new expanded surround focus mode. In addition to the keepers, I have quite a number of shots of perfectly still birds that are soft (front or back focused). The camera nailed BIF but was a little less dependable for static scenes in the settings I used. I wish I has shot a few of the static birds in One Shot to see if focus was more consistent but I didn't.

Overall, I'm happy with the body, but can't quite understand why this focus issue exists for good contrast static subjects. It seems like a firmware tweak is in order.

Didn't they update the 1DX Firmware to fixup AF issues/inconsistencies​? It's very possible they could.
But since it doesn't seem to be a wide spread issue, I'm still not sure.

Unless, we are being very observant/critical and others brush it off as user error or don't care?
Something most definitely goes wrong, and I feel it's gotta be the hardware/new methods being used.

I bet when we send our stuff into Canon, they test only in OneShot. But I'd hope they test in AI Servo someway or another.

I noticed in my settings after getting mine back, they were shooting in JPG with Noise Reduction On and some other things turned back on.
But they did not reset to factory settings or anything.

I think other than critical photographers, most people look at the overall pic and say "Good enough". Their good enough, my good enough and your good enough can be all different.

I exchanged my 3rd body today and moved to a fourth. Showed the samples to my local store and they were pretty floored.

huntersdad wrote in post #17416372
... Kinda of in a quandary - exchange it and start over or send the 7d2/600/1.4/2x all in and let Canon sort it out. They've had it once with no appreciable better results.

There is no guarantee that they will fix it. I sent my Canon 17-55mm lens in for AF malfunction and after 4 trips, it still was not fixed. Sent it in again and am awaiting the results of the fifth attempt.

One problem with Canon service is that they don't seem to read the complaint very carefully. My lens focused badly at far distances. So what do they do? Check focus on nearby objects like bottles, find no problem, make a minute adjustment so they've done something, and send it back.

YMMV, but you might also be hassling with Canon for a long while.

Maybe ask them if they would pay rental for your valuable gear.

huntersdad wrote in post #17416891
I think other than critical photographers, most people look at the overall pic and say "Good enough". Their good enough, my good enough and your good enough can be all different.

I exchanged my 3rd body today and moved to a fourth. Showed the samples to my local store and they were pretty floored.

Hope the 4th is the one. My window passed last week. My hit them up again tho as I had alerted them of issues...

You're on your 4th body? wow.

If we are having these issues, I can't imagine what people think who bought the 7D II as their first DSLR.
Which I saw first hand some people looking to do, at the local shop.

I saw like 2-3 people with 7D mark IIs last weekend. Next time I see them, i gotta do some comparisons or swap bodies and see.

Yep. First 3 with all different issues - body 1 wouldn't focus period, body 2 wouldn't and would misfocus on things sitting still, body 3 did better on the tracking but randomly front focused or back focused by amounts that couldn't corrected with MA.

So, here we are at #4. We'll see how it does at anything.

Inconsistent front / back focus in Ai Servo issues....
In the spirit of trying to help and understand:

I have a new theory regarding these issues. I believe they are related to the poor detail on the impact of the underlying Case settings. In the past, the line of thinking was to set Tracking to - 2 (slowest to change focus area) when doing focus testing in Servo mode, and with recommended Single Point or Spot focus to, aiming known behavior and minimize camera auto selection etc.

I think that some of reason could possible be initial focus lock area in the scene as you move the lens to lock on to desired target and compose the shot. Might it be that despite the focus point square being enveloped by subject after the shot, is it possible that it locked on to where the sharpest focus occurred based on acquisition and continued through the shot to stay locked despite final composing? . I read somewhere that a successful users Top recommendations was to make sure you were properly tracking for an exaggerated 1 second to make sure it was locked on Your target.

Maybe increasing Tacking a little would have similar benefit. The 7dii seems very fast and exaggerated algorithms compared to other bodies. Track might be much slower to than other bodies with a - 2 value. . Anyone care to try these techniques and see if improve, eat occurs? Of course this assumes MFA completed, and redone if TC is in use. If not, all bets are off.

huntersdad wrote in post #17417116
Yep. First 3 with all different issues - body 1 wouldn't focus period, body 2 wouldn't and would misfocus on things sitting still, body 3 did better on the tracking but randomly front focused or back focused by amounts that couldn't corrected with MA.

So, here we are at #4. We'll see how it does at anything.

4 7D2s? What is the retailer you're getting them from?

I'm just asking so I don't use them if I decide to pick one up.

I went through 3 bodies and have been keeping detailed records of my hit/miss ratio with the 7D2...and have concluded that while there are some funny things going on now and then, this body can focus and I can get keepers 85%-95% of the time. I also know that my 5D3 can achieve a similar keeper rate...so I've got enough data to conclude that my experiences with my first two bodies weren't my fault, and that my current body can focus consistently. I recently posted about these things in this thread a couple pages back...charts and graphs and all. My only remaining concern is the hit/miss ratio in Case1 using AF Point Expansion modes...which don't appear to perform as well as some of the other combos.

For those still having troubles it would be helpful to understand the scope of these focus issues in the larger context of keeper rate. I don't ask this question as a criticism: does anyone with a problematic 7D2 have any hard data on their hit/miss ratio? Anecdotal information is not "evidence" and establishing a baseline performance profile (using a known good camera with a known good lens) is important as point of reference for assessing the performance of a new body like the 7D2...if it's your first copy or your 4th (sorry to hear, huntersdad!)

Anyways, Im just making an argument that hard data is invaluable...

thx
PP

ppmax wrote in post #17417399
For those still having troubles it would be helpful to understand the scope of these focus issues in the larger context of keeper rate. I don't ask this question as a criticism: does anyone with a problematic 7D2 have any hard data on their hit/miss ratio? Anecdotal information is not "evidence" and establishing a baseline performance profile (using a known good camera with a known good lens) is important as point of reference for assessing the performance of a new body like the 7D2...if it's your first copy or your 4th (sorry to hear, huntersdad!)

Hard data and stats are nice, but when one shoots subjects 100 yards away then 20 yards, some flying, and some stationary and some running. Some with 24-105, some with 100-400 and some with 400 DO, etc. It's hard to create a statistically sound data set. Just too much statistical noise shooting wildlife.

I don't think the hard part is collecting data...the hard work is in categorizing it and analyzing it. Any modern photo editing app will let you create queries or smart folders or filters to create buckets of photos with similar characteristics. Once you have buckets of similarly shot photos, you can grade them via stars or labels or whatever according to a simple set of rules (in focus, near miss, total miss). Once you've done this you can count how many keepers you get in each bucket. Transposing this info to a spreadsheet is time consuming, but not hard.

For those that are interested here's my workflow:
Import pictures to a folder
Run exiftool to rename the pictures by date-time-subsecond like so: 2014-02-04_12-15-30-05.cr2. Renaming like this makes sure that your high speed sequences are always sortable by the order in which they were taken.
Run exiftool again to add keywords for which Case and which AF Area Mode was used. Any shot may contain keywords Case 2, AF Expansion (8pt)
I then import these images into Aperture.
Once the import finishes, I label each photo with a color: Red=total miss, Yellow=near miss but acceptable at 25% or 50% zoom, Green=nailed it.
After labeling, I create filters for each case...and then sub filters for each AF Area Mode.
I then sort the display according to label. This enables you to see all the Red, Yellow, and Green pics in groups; count the items in each group.
Enter this data into a spreadsheet. Total up each row, column, etc...charts, graphs follow.

IMAGE LINK: https://flic.kr/p/qK5m​Tg

gschlact wrote in post #17417303
Inconsistent front / back focus in Ai Servo issues....
In the spirit of trying to help and understand:

I have a new theory regarding these issues. I believe they are related to the poor detail on the impact of the underlying Case settings. In the past, the line of thinking was to set Tracking to - 2 (slowest to change focus area) when doing focus testing in Servo mode, and with recommended Single Point or Spot focus to, aiming known behavior and minimize camera auto selection etc.

I think that some of reason could possible be initial focus lock area in the scene as you move the lens to lock on to desired target and compose the shot. Might it be that despite the focus point square being enveloped by subject after the shot, is it possible that it locked on to where the sharpest focus occurred based on acquisition and continued through the shot to stay locked despite final composing? . I read somewhere that a successful users Top recommendations was to make sure you were properly tracking for an exaggerated 1 second to make sure it was locked on Your target.

Maybe increasing Tacking a little would have similar benefit. The 7dii seems very fast and exaggerated algorithms compared to other bodies. Track might be much slower to than other bodies with a - 2 value. . Anyone care to try these techniques and see if improve, eat occurs? Of course this assumes MFA completed, and redone if TC is in use. If not, all bets are off.

I'll test them next time I'm out, which will be next weekend. There's a train of thought developing that the MkII superteles are already a little "hyper" on the focusing, meaning it seems that they are trying too hard sometimes. Combine that with a camera with speed in focusing and "hyper" tracking and you would have problems.

ppmax wrote in post #17417399
I went through 3 bodies and have been keeping detailed records of my hit/miss ratio with the 7D2...and have concluded that while there are some funny things going on now and then, this body can focus and I can get keepers 85%-95% of the time. I also know that my 5D3 can achieve a similar keeper rate...so I've got enough data to conclude that my experiences with my first two bodies weren't my fault, and that my current body can focus consistently. I recently posted about these things in this thread a couple pages back...charts and graphs and all. My only remaining concern is the hit/miss ratio in Case1 using AF Point Expansion modes...which don't appear to perform as well as some of the other combos.

For those still having troubles it would be helpful to understand the scope of these focus issues in the larger context of keeper rate. I don't ask this question as a criticism: does anyone with a problematic 7D2 have any hard data on their hit/miss ratio? Anecdotal information is not "evidence" and establishing a baseline performance profile (using a known good camera with a known good lens) is important as point of reference for assessing the performance of a new body like the 7D2...if it's your first copy or your 4th (sorry to hear, huntersdad!)

Anyways, Im just making an argument that hard data is invaluable...

thx
PP

I have not kept stats as they would be so bad at this point it wouldn't even be funny. My keeper rate with the camera is well under 10%, I'm guessing under 5%. And that's being generous in keeping some shots that I might would otherwise delete. I will say that when the combo performs, it is knock out good - colors are good, sharpness is excellent, noise is well controlled. But when it doesn't it is infuriating and frustrating.

As a side note, I did hear back from the CPS rep who reviewed my files and said there was definitely an issue with focusing. He took the serial number of my 600II and 1.4x as they have now had 2 CPS members with major problems with that combo (pretty sure I know who the other is). Since I exchanged it, we're going to repeat the process - shoot, send pics, let them review and go from there. He's hoping to avoid me having to send my 600 in and hoping they are able to compile some data and determine if there is a problem within a sect of serial numbers of the lens or TCs or if there is just something in the camera that's off. They have tested the body with a 600I and TCs (v. II and III) and have had no problems whatsoever.

ppmax, the hard part would be making numbers between, say, me and you, consistent. What would be a near miss for you, might be a full miss for me. Something that simple could skew my numbers vs. yours and make your camera look better than mine, when in actuality, it could be opposite or performing the same. I shoot for a publication and can guarantee you I delete shots 99% of people would keep, with the exception of a shot that I think is so good that I'll overlook the flaw, which is truly rare.