AFMA results - Conflicting values

[Steph]

I've AFMA-adjusted my 70-200mm f/2.8L II IS last week-end but I'm a bit puzzled by the results I got. Of course, I made the tests several times for every focal length but, at least, those are consistent.

Focal gives me the following adjustments:
- 70mm: -2
- 100mm: 6
- 135mm: -1
- 200mm: 6

OK, now, what do I do with that... Even a 1D X with the double AFMA values wouldn't help here.

I also saw that the Quality of Focus is better in the middle of the range but that's known that the zooms are worse at the extremes, isn't it.

Any advice??

[Steph]

For those interested, here are four graphs showing the measurements for the different focal lengths



Sorry for the JPEG quality but I'm limited to 195 kB.

[cls]

Hi, if you're convinced the lens/camera combo doesn't focus properly, I would say you should let Canon adjust it. I don't know the warranty policy in your country, but at least where I live (Sweden), such calibration is included in the camera body warranty (for fairly new lenses). AFMA can partially compensate for certain deviations, but as it seems not in your case. By the way, did you test at different distances or only at some single distance? That can also affect the result.

[HDNitehawk]

What camera are you using? What are you testing with?

The results you posted are not too consistent.

I had varying results as well with my 70-200mm, it was more noticeable on the crop cameras than it was on the 5D II. Mine however was more consistent. It was off opposite directions on the very ends and had very flat results through the range. It never at any point was so far out that it was outside a reasonable DOF.

Really I think you would want the zoom to be its most accurate on the long end, because of your shallow dof. The short end is more forgiving.

[Steph]

What camera are you using? What are you testing with?

The results you posted are not too consistent.

I had varying results as well with my 70-200mm, it was more noticeable on the crop cameras than it was on the 5D II. Mine however was more consistent. It was off opposite directions on the very ends and had very flat results through the range. It never at any point was so far out that it was outside a reasonable DOF.

Really I think you would want the zoom to be its most accurate on the long end, because of your shallow dof. The short end is more forgiving.

I'm testing with a 7D on a tripod 4-5 meters to the target. I'm using the latest version of Focal. The results for a given focal length are actually quite consistent and reproducible so I guess the values given are actually the best for that length.

You're right DOF will be smaller at 200mm so I should probably pick up something closer to 6 than -2.

[neuroanatomist]

Those 'peaks' are pretty broad - mine were much sharper on both the 5DII and 7D (see graphs). When I see inconsistent results with FoCal, my first question is lighting - what were the typical EV's reported for your testing? Mine are always in the 11-12 EV range, and I get consistent results with most lenses.

[HDNitehawk]

On my crop bodies (7D and 1D IV) the variance was exaggerated more than it was on the 5D II.

Your results concern me because they go up and down over the range. Mine were very consistent as I went up to 100mm - 135mm and so on. Only on the very ends did I get a sharp spike one way or the other. I would make sure that the madness isn't in how I tested first, which is often the case. Then if I kept getting the same results I would send it to Canon for service.

[Steph]

Those 'peaks' are pretty broad - mine were much sharper on both the 5DII and 7D (see graphs). When I see inconsistent results with FoCal, my first question is lighting - what were the typical EV's reported for your testing? Mine are always in the 11-12 EV range, and I get consistent results with most lenses.

Good point! I'm not that "bright" 7.8 EV from a 500W spotlight.

[neuroanatomist]

Good point! I'm not that "bright" 7.8 EV from a 500W spotlight.

When I read the manual initially, it recommended having at least 10 EV. Low light leads to inconsistent results in part due to the longer shutter speed required (vibration reducing sharpness, but variably). Vibration can even be an issue with shorter shutter speeds. When I started testing, even with my 12 EV, I'd get failed runs sometimes - the setup was on the main floor of my house, hardwood floors, and me or my kids walking around was the issue. I moved the setup into the basement (concrete slab floor) and the testing was much more reliable.

I use three 150 W gooseneck lamps, so I actually use less total wattage. But...remember the inverse square law. My lights are positioned about 12" from the target.

I have a question, or two, about the Focal software. Does it step through the adjustment range while it is running to find where the image is sharpest? Reason I ask is that I see in the graphs that it covers such a wide range that I wouldn't expect to see from a camera body. The other question is what do the numbers of the Y-axis actually represent as the measured quantity?

Yes, it iteratively steps around the range from -20 to +20, and uses predictive algorithms to refine the values tested. The plots I posted are from a very early version of FoCal (IIRC, the first version to offer reports because I have just screenshots from older versions) - I would not read anything into the Y axis. But newer versions offer an image quality test, so I suspect the numbers may be more meaningful now. Even with my older tests, it's no real surprise that the 5DII has higher Y-axis values than the 7D, right?

[Steph]

I use three 150 W gooseneck lamps, so I actually use less total wattage. But...remember the inverse square law. My lights are positioned about 12" from the target.

Alright, I'll try to setup something in my garage then with the 500W spotlight reeeally close to the target. I'll post other results and let you know how it went.