5D3: AFMA variations between different AF points

mco_970 wrote in post #15278114
Michelle - how much light do you have on your test chart? It should say how many EV's you have during testing. More EV's will make your resolution #'s better. Could be that you need more light. Also be sure to save the test results and look through them. I noticed on one of my test charts that I was having a glare issue, and it affected results. I had to change to a matte paper instead of luster and re-run the tests to make sure the results were correct.

My 50/1.4 had numbers in the 900-1000 range with ~9.5 EV light, and better than that once I moved my setup where I had closer to ~12-13 EV's based on natural light.

on the last test it said my EV ranged from 6.8 - 7, so I guess i will have to move rooms as that is as much light as i can get in the room where my computer is set up.. I printed the charts just on plain white paper, but have another blank white piece of paper underneath it so that the white is bright, they are only on A4 size though - is that big enough? On the tests the green tick came up so I thought they would be ok and on the multifocus I did have to move the chart closer than the other chart for it all to be on the frame, but got the green tick so thought it was ok... maybe that one needs to be printed bigger? going to be a big effort but I will move everything to a different room tomorrow, retry and tests the 100mm too. Thanks for all the help so far! really appreciate it!

michelle blossom tree wrote in post #15278785
on the last test it said my EV ranged from 6.8 - 7, so I guess i will have to move rooms as that is as much light as i can get in the room where my computer is set up.. I printed the charts just on plain white paper, but have another blank white piece of paper underneath it so that the white is bright, they are only on A4 size though - is that big enough? On the tests the green tick came up so I thought they would be ok and on the multifocus I did have to move the chart closer than the other chart for it all to be on the frame, but got the green tick so thought it was ok... maybe that one needs to be printed bigger? going to be a big effort but I will move everything to a different room tomorrow, retry and tests the 100mm too. Thanks for all the help so far! really appreciate it!

A4 or 8.5x11 is what you want for the main AFMA test chart and aperture test. For the AF point test you need to print the other chart on multiple pieces of paper and tape them together. I ended up with 3 sheets wide and 2 tall which enabled me to be between 25 and 50x from the target.

Need. More. Light. For a few bucks/quid you should be able to nuke the chart to acceptable brightness. As you mentioned A4 paper you may not not US based? If US, go to Home Depot and grab 1000 watts of halogen work lights for about $35 and you can light the entire neighborhood and fry some eggs on it when you are done.

Yep, you definitely need more light. I have to borrow a laptop to run Focal, so I understand the pain of not being able to run it where you want!

Definitely have a chart printed on matte paper. It will be higher resolution, lower glare, and you'll get better results. You should be able to do this at Costco or Walmart or whatever, or just buy a small pack of matte paper. Plain paper is disrecommended by FoCal. My target is 8.5 x 11, but I use several of them side by side for the focus points check so that I can move my camera back to a reasonable distance.

Basically - make your target setup as perfect as you can. The more time/effort you spend at the start getting that part right, the better your results will be.

s2kdriver80 wrote in post #15275028
Ok, so I sent in both of my 5D3s and my 24-105 last Thursday, arrived at Canon NJ Friday morning. Monday I call the NJ office to see how the repairs were going and when I could expect them back since it would be great to have them in time for Thanksgiving. Fools didn't even open the box yet and it was still sitting at their warehouse since Friday morning. What's the point of placing the CPS decal on the outside of the box for express service as indicated in their instructions if my shipment falls through the cracks and gets left in their receiving bay for an entire day doing nothing? Anyway whatever, I let it slide. I figured if they fix the AF issues, what's a little delay right?

So as Tuesday rolls around, it becomes evident that I won't be receiving my gear in time for the holidays. I called them and offered to drive all the way from Long Island to pick them up in NJ on Wednesday, even though the CPS return shipping is free. I fought the worst traffic of the year and finally arrived at the NJ office. I went 100 miles in 4 hours, driving manual in bumper to bumper gridlock.

The lady returned my gear and the work order forms. "Electrical adjustments were carried out on the af assembly. Product functions were confirmed. Checked all, adjusted focus & exposure, f&b, cleaned to factory specs. Cleaned C-mos & adjusted with lens." That's it. I asked if I could have the tech notes and she said she didn't have them. "Electrical adjustments" are meaningless to me lol. I want to know which AF points were adjusted and by how much and whatever else was done. This way, I would know if the tech was on the right track and if my issue ever needs to be revisited by that tech or another tech, they would know the history of the work that was done. Surely these techs keep notes, right?

So I brave the traffic back to Long Island and was eager to run the bodies through some basic testing. I set up the FoCal target, 12EV halogen brightness, 40x focal length, tested 105mm at f/4. Center point AFMA now yields +5 wtf lol! Whatever, if all of the AF points are around +5, I'll live with it. I then tested the upper-right-most outer cross point. +11. In other words, not only is my center point a +5 now (was 0 before), there is now at least a 6 AFMA units of difference between my center point and the outer point I mentioned - more than before.

I stopped the testing right then and there, no point in continuing with other AF points or with the second body. These tests are time-consuming and I don't think I should be the one doing this after receiving a "fixed" product. This is the reason I want to see all of the tech's notes, including what was done to fix the AF and also their findings and observations during the quality check. Since this is free warranty work, do you think these guys do haphazard, non-involved work and hope the customer just lives with the issues and pray they don't come back a second time? Or did I just get a bad tech? Just trying to fathom how this could have happened. Seeing such poor results, I just don't see how the tech could have signed off on the quality check. Doing the "electrical adjustments" implies and is an admission that he found some fault with the AF. So the result is worse and he is ok with it and verifies the work to be handed back to the customer? I don't get it. Now I have to waste more time and more money returning my gear to the repair center again. Any word of advice on what/how I should phrase my repair form?

Anyway, Happy Thanksgiving to all who celebrate. In the grand scheme of things, these cameras are insignificant compared to other more important things in life and we should be constantly thankful for what we have, just a little annoyed that they half-assed the work when it didn't need to be that way.

Paul - Sorry to see you having so much trouble! I had similar AF difficulties with a Rebel XSi/450D + 100-400L. Had to send them back 3 times! The way that I finally got it resolved was to call 1-800-OKCANON and told them that I wanted to speak to a supervisor in Virginia (location where my work was done). He got involved personally and finally got it fixed.

Good luck!

Yogi

Thanks for all of the feedback guys, appreciate it!

A subsequent visit did not improve things so I asked them to just delete the offsets they had made so that it would be back to factory condition. Even though the AFMA numbers weren't the exact same for every AF point, the numbers were better out of the box. Lesson learned - unless you have a clear defect or a malfunctioning camera, do not send your gear to Canon, especially if the "fix" has a somewhat subjective component. Those guys were friendly, accommodating, and willing to do whatever it took, but somehow the adjustments made just didn't jive with AFMA accuracy in real world scenarios. Like someone had mentioned earlier in this thread, I'm thinking this was as-designed and functioning perfectly normally, I was unrealistic. After they reset the numbers back to factory settings, it's much better and how it should be. I've also realized these AFMA variations between AF points has a lot to do with the lens you use. My 70-200 is nearly bang on perfect on every AF point!

I think many of us are having the same problems.. Really fed up having to use the 5 centre points all the time. After hours and hours on FoCal, I am just going to bite the bullet and send it to canon. Hopefully they can figure it out.

Some lenses perform more consistently than others too.

I take AFMA readings with FoCal from 5 AF points (center and the four outermost cross-type corners), then sort of average them. If the AFMA values are within say 3 points, I wouldn't sweat it, no body/lens combo will give the exact same AFMA value for all 61 points.

s2kdriver80 wrote in post #15642576
Some lenses perform more consistently than others too.

I take AFMA readings with FoCal from 5 AF points (center and the four outermost cross-type corners), then sort of average them. If the AFMA values are within say 3 points, I wouldn't sweat it, no body/lens combo will give the exact same AFMA value for all 61 points.

Thanks for the tip. I will try that tomorrow.

And make sure you have enough light shining on the target. Enable off-centre points in the settings, so that FoCal will let you choose non-centre points. And when you test the outer points, make sure to actually move the tripod to the left or right so that you are perpendicular to the target (not just turn the ballhead so that the AF point is on the target).

The right hand points on my 5D3 were about 10 MA points out from the centre and left hand points. I've sent it in, the technician called me to tell me that the outer points were 'within canon specs', I asked how far out they were from the center on their focus chart and he said 'just a few cm'. This with a 50L wide open. In any case he now tells me the right and centre are equal, I hope that is equal for real rather than 'within a few cm'. Sheesh.

s2kdriver80 wrote in post #15199317
Before a recent wedding shoot (wasn't the main photographer), I had calibrated my lenses and bodies, including a 5D3 and the 24-105 lens with the center point. In most of my shots, there is a slight front focus in each of them. In upper body and head shots, the focus is good or the OOF is very close to the intended point of focus so the photos are acceptable. For full body shots (where longer camera to subject distances exaggerate AF misfocus errors), most of the photos are unacceptable. AF'ing on someone's face rendered focus on the hands resting in front of them, or their shoulder if they were standing at an angle.

Realizing that I used the outer cross points most of the time, thought had occurred to me to try comparing the AFMA values for the center (which I had already calibrated correctly) and the outer cross points I'd used. Using Reikan's FoCal tool, I consistently get around a -3 AFMA setting difference when comparing the center to a couple of outer cross points, which makes sense since they were causing front focus. Then using FoCal's multi point focus analysis tool, the results show those outer points around 75% of the focus quality of the center point (when using an AFMA value calibrated for the center point).

Since AFMA adjustments are global and can't be adjusted per AF point by us, is this something that a Canon repair facility can fix? OOF from an f/4 lens was noticeable, it would have been even worse had I used a faster lens. I don't mind if all of the AF points as a whole were off 5 or 10 AFMA settings to a lens, since I can adjust. But if there are AFMA variations between different AF points, how can some points be trusted? Has anyone had a similar experience where different AF points behaved differently? The AF systems in the 1DX and the 5D3 are supposed to be Canon's best and the costs reflect, am I unreasonable in expecting a tighter tolerance?

Just wondered if you have hit an issue I am looking into now. That being the degradation of AF point sensitivity with narrow maximum aperture lenses. Eg the 1DsMk 3 has 4 break points of max apertures of lenses :-
f2.8, F4 , F5.6 and F8. I.e if the max aperture of a lens is greater than F2.8 then the 9 centre AF points are highly sensitive cross hairs horizontal and vertical. Yet with a lens of max aperture of F8 you are down to just one single centre AF point and that is now not even cross sensitive - just horizontally sensitive!
Your using an f4 lens if I am not mistaken - so by then on a 1Dsmk3 you'd see a massive degradation.
Does the 5D3 have this "feature"! Especially can cause problems if you are still using outer AF points.
I could not believe what I was reading at first because you fundamentally needed f2.8 constant lenses (which of course a lot of pros have - so it's not an issue) but for the serious amateur this could be a real deal breaker, or worse lead to unexplained problems.

So I did my test first with my 85L today, repeating the test three times on each AF point (top, top and bottom right, top and top left).

Centre -1, -1, -2
Top right -4, -6, -4
Bottom right -4, -3, -3
Bottom left 5+, 3+, 5+
Top left 1+, -1, 0

I did the same with my 200L and it seems very similar with the 85L, with the right side consistently around 7-9 AF points different to the left. It's as if the AF plane is slanted or something. I think I need to send it to canon and hope they can figure it out.

Hi all,

I've had to register just to post to this discussion.

I've had my 5D3 since Nov '12 and I've noticed inconsistent AF previously and chalked it down to user error. However, just last w/end, doing a model shoot for several hours for the first time with the 5D3, using singe AF point located in the middle of the 2nd vertical row from the right and holding the camera in portrait mode, my keeper rate was well below 50%.

I was taking almost 5 shots for each pose and I was lucky to get 1 shot extremely shart and 1 shot with acceptable sharpness. The remaining 3 shots were junk. This was with the 85 f/1.8 and 135 f/2.8 SF (admittedly not the best lens but neverthe less has worked well before).

This got me really intrigued and I did some tests with camera mounted on a tripod shooting a newpaper on 45 degree angle with 3 different lenses - 85 f/1.8, 135 f/2.8 SF and 100L f/2.8 macro. I have noticed that the AF points on the right side is abysmal, with a sharpness rate of around 50% and the centre AF point is roughly 75-80%. To double check, my 5D2's centre point could easily achieve 80+% of shots in focus.

I've been searching high and low for reported problems with 5D3's AF consistency (particularly the outermost points on the right as they are most useful for portraiture) and no one seems to have reported it until I came across this discussion.

Glad to hear others have the same issue but disappointed to hear there doesn't seem to be a guaranteed solution.

I shall be using FoCal this w/end to do more testing but I think my problem is the same as that experienced here.

I haven't tried (not at home) and I'm sure it can't be done, but can different MFA's be done for different custom settings for the C1-C3?

Would be awesome if it did, or at least Canon updated firmware so the C1-C3 can store different MFA. Then if you decide you want to use outer focus points, you can switch to one of your customs which has the MFA ideal for that. Back to center focus, switch out of it, etc. Not ideal, but better than trying to switch the MFA manually on the fly!

I'm sure the Custom settings won't store different MFA's, but people should write to Canon to demand a firmware update so it can.

Someone just opened a thread at the Magic Lantern forums, asking for a new feature that allows to set AFMA for specific AF point(s).

http://www.magiclanter​n.fm/forum/index.php?t​opic=4966.0

Jump in and give your support! The more the merrier.

This is something that Canon and / or Magic Lantern should have done already.