5D III and 300 2.8 II hit rate for sharp pictures very low

[Flo1111]

Good morning,

Yes you are all right I did not use professional tools for the AFMA.

I put a wooden foot rule on the floor. The lens on a tripod, about 3 meters away. DOF was about 1 cm.
Now I did focus on one line e.g. at 20 centimeter.
In that situation I did realize that the focus was behind, the picture was more sharp at 21cm. (I did not only focus once, but several times. I alway but something in between and focused on that after wards again on the foot ruler. The cam and the foot ruler I never did move)

After about two or three hours and amy photos I though I am ready. The AFMA for that lens was at -20. (the maximum)
I already was very surprised, because only at the far end the dash for the 20cm was really sharp.

After that I also download on test graphic from http://www.the-digital-picture.com, which was mentioned here in the forum.
Put the graphic on the wall and straight to it the lens. But now everything was completely unsharp. Also for real photos that setting was not correct. So I set everything back to 0.
It just does not fit together. :-(

I think I should buy the Software mentioned here and give it a try.

By the way. The test photos that I posted here from my window, the the solar system is about 40 meters away, what should give me a DOF of 2,5 meters. In my opinion I should get 9 from 10 photos very sharp.

Thank you.
Bye
Flo

[AlanF]

Two big mistakes here. First, Canon recommends you test focus at a distance of 50 times focal length between lens and target (ie 15 m for a 300 mm lens). Second, if you use a home made rig, then the target should be vertical and perpendicular to the lens, and the ruler placed along side it at say an angle of 15 degrees. Then focus on the target, not on the ruler and note which line on the ruler is sharpest. I do the test by having a black line on a sheet of white paper pasted on a cornflakes box or similar and propping a ruler next to it at the same height as my camera on its tripod. Try that. 

Being an obsessive scientist, I plot the mean of each series of measurements at a particular MA value versus the MA value and draw a line through them to read off the graph where the MA value gives the right focus.  But that is not necessary.

[neuroanatomist]

I put a wooden foot rule on the floor. The lens on a tripod, about 3 meters away. DOF was about 1 cm.
Now I did focus on one line e.g. at 20 centimeter.

If I had $1 for every time I've read about someone using that 'method' I'd probably have a 300/2.8 II to go along with my 600/4 II. 

Here's the thing - the AF system is going to lock onto the feature of highest contrast at the correct orientation under the selected AF point. Note that the actual AF point is larger than the little box representing it in the VF (even with Spot AF, it's still slightly larger, although you should not use Spot AF for AFMA).  So, you know that you intended to focus on the thin 20cm mark running horizontally, but your camera saw the higher-contrast feature running vertically across the point - the edge of the ruler.  That feature extends away from the camera, and anywhere along that edge would be 'correct' for the AF system.  No wonder your AFMA 'made things worse'.  That's why the commercial tools use a flat target oriented parallel to the image plane.  Also, you were far too close - AFMA should be done at 25-50x the focal length of the lens.

The best test to see if AFMA is the solution is what AlanF alluded to a few posts back - set up on a tripod, with a flat, high contrast target (a sheet of newspaper taped to a wall, for example).  Activate Live View, and take several shots using Live AF each time (not Quick AF - you should not hear the mirror flip during AF, you want to be using contrast detect AF where the CMOS image sensor is used to focus). Then, turn off Live View and take several shots with 'normal' (phase detect) AF. If the Live View shots are sharp and the regular AF shots are not, AFMA should fix that (and I really do recommend Reikan FoCal, in that case). If the Live View shots are soft, it's likely a problem with the lens, probably meaning a trip to Canon Service.

Hope that helps...

[rpt]

Hi,
I'd like to hijack this thread. I have the same problem IMHO with a 5D MKIII and a 70-200 F2:8 L IS II USM that I just bought. I have the feeling that I never get the sharpness that I expected when using AF so this may be just a problem of mine. I've posted a few times on CR about my problems, so I wanted some feedback before sending the Camera and Lens to Canon for checking out
I shot about 100 pics using AI Servo today of my Dog Dave, and our lodger Timmy. Most of them the dogs were walking slowly or trotting, so no high speed motion. Light was good, and I guessed that 1/500th should do, and I was using Timer priority. Spot metering. Using Case 1 AF as the dogs are pretty predictable. Zone AF.
As I had the same problems yesterday I did an AFMA on the lens yesterday evening, but I saw no improvement today.
As stated here already,  I too expect this lens to perform OK at 2.8.
FW was at 1.1.2 for these pictures, but I just updated to 1.1.3 just in case.
So looking at this example IMHO I still have a serious back-focus problem on this lens.
Anything else wrong with my setting?

Cheers Brian

Yes, it seems to be back focused. How did you AFMA it? I have had major errors during AFMA and was frustrated but they all turned out to be user error - tripod not stable and bad lighting... Once those were fixed, no issues. I use FoCal on my 5D3.

[Flo1111]

Hi AlanF
Hi John,

got it. I made everything wrong what I could do wrong.
John I did read again your article http://www.the-digital-picture.com/Photography-Tips/AF-Microadjustment-Tips.aspx

Ok so I will try now the live view / normal AF comparison.
I will us the the png file from your article

Thank you again very much for your help.

Bye
Flo

[bjd]

Interesting infos about the 1DX Firmware.

You can download the software from the Canon website here.

Firmware Version 1.0.6 incorporates the following fixes.

1. Fixes a phenomenon in which the AF Microadjustment cannot be correctly carried out.

2. Fixes a phenomenon such that when using the camera with combinations of certain lens*1) and extender*2), the images captured become back-focused.

*1) EF 70-200mm f/2.8L IS II USM or EF200mm f/2L IS USM
*2) Extender EF 1.4x III or Extender EF 2x III

IIRC up till now almost everything concerning AF on the 1Dx also applied to the 5D3.
Been playing with Focal all morning without any improvements.
Cheers Brian

[rpt]

Interesting infos about the 1DX Firmware.

You can download the software from the Canon website here.

Firmware Version 1.0.6 incorporates the following fixes.

1. Fixes a phenomenon in which the AF Microadjustment cannot be correctly carried out.

2. Fixes a phenomenon such that when using the camera with combinations of certain lens*1) and extender*2), the images captured become back-focused.

*1) EF 70-200mm f/2.8L IS II USM or EF200mm f/2L IS USM
*2) Extender EF 1.4x III or Extender EF 2x III

IIRC up till now almost everything concerning AF on the 1Dx also applied to the 5D3.
Been playing with Focal all morning without any improvements.
Cheers Brian

What firmware version are you on? The site lists 1.1.1 as the latest version...

[neuroanatomist]

I think his point was that the 5DIII may have the same problem, which was fixed for the 1DX but not for the 5DIII.

[rpt]

I think his point was that the 5DIII may have the same problem, which was fixed for the 1DX but not for the 5DIII.

Right! I don't have those lenses nor the extenders...

[bjd]

Interesting infos about the 1DX Firmware.
..

What firmware version are you on? The site lists 1.1.1 as the latest version...

Neuroanatomist is correct in what I meant. 
I'm on 1.1.3 (the latest for the 5D3) now.
Also I wonder if they really can be certain that the problem does not occur on other lenses too?
So, yes,  I'm hoping for a new 5D3 FW too sometime.
Cheers Brian

[rpt]

Interesting infos about the 1DX Firmware.
..

What firmware version are you on? The site lists 1.1.1 as the latest version...

Neuroanatomist is correct in what I meant. 
I'm on 1.1.3 (the latest for the 5D3) now.
Also I wonder if they really can be certain that the problem does not occur on other lenses too?
So, yes,  I'm hoping for a new 5D3 FW too sometime.
Cheers Brian

I have not had any issue with my 24-105, 40mm, 100L macro or the 100-400L.

[revup67]

Hi BJD - I had read this article on AI Servo settings a few days ago wanting to get even more in focus BIF shots.  That 5DM3 has a lot of options for tweaking AI Servo.  If you've not gone through these setings (all 6 cases with 3 settings on each) then this article may help you:

http://www.garyluhm.net/bio/tips_0512.html

[Flo1111]

Hello neuroanatomist,

I did do the normal AF, live view comparison and the life view pictures where alway sharp the normal AF photos something between.

I did do it in that way. Pu the graphic on the wall with some additional light and about 7-8 meter away, on the same level, the lens on a tripod. (Thats the longest distance that I could find.)

So it seems that AFMA could help me out.
I think I will buy FoCal.
What version would I need? Plus or Pro. What would you suggest?

@bjd yes you are right it could be, but Canon would never tell anything. Perhaps in April everything will be better.

Thank you very much.
Bye Flo

[neuroanatomist]

I have the Pro version, but the Plus would be fine for the 5DIII. I do like the additional tests with the Pro version (testing all the AF points, etc.).

[mingyuansung]

Flo1111.  Please let us know the result if you ever bring your camera and lens to Canon Service Center. I really want to  know if they can help. By the way, which center you are going to? Thanks.