@ebiggs1 wrote:"I have not heard of this complaint before with ANY of the Canon 70-200mm lenses."
I totally agree with this. Although highly unlikely to be the problem with that lens it is not impossible. What you need to do is set up a way to measure it not just shoot at a flat plain. A yard stick for instance. On a tripod focus at a midway point on the ruler and see if focus was exactly at the place you aimed the camera. The camera/lens must be steady. Use just the center focus point. Use the cross-type focus point. Use One shot, never Ai-servo mode. Turn off all lens corrections in your camera.
"I used a tripod and aimed directly on the eye."
If you used multiply focus points when you took the photo of the "eye" and perhaps Ai-servo focus mode that is a better cause of your problem.
When I have tested focus, I use either a ruler or a yardstick at an angle, combined with a vertical target for focusing. Which setup I use depends on the focal length I am testing. The red boxes are red bricks with a yardstick target.
I wouldn't consider having to slightly calibrate the AF a complaint with any lens. There is going to be some variance from lens to lens and camera body to body so sometimes the variance of the combined pair is going to be of sufficient direction and magnitude to create an issue WHEN the depth of field is extremely shallow.
I pretty much follow the Canon adjustment guide I linked earlier except I do a series of shots at varying calibration offsets in one run so that I can quickly choose either the correct offset from this first run or know which two offset points to choose between to make a second run. Typically when dialing in a lens and body I will shoot three each at 0, +3, +5, +8 -3, -5, -8 and then carefully check all of these with a computer and large monitor. It is very important to use a single focus point for all of these tests and to use either a tripod or sufficient shutter speed to ensure that camera motion blur doesn't impact the results. You MUST follow Canon's direction to take multiple shots at each offset and manually throw the focus off between to ensure valid results.
I don't use any special target, for example when I dialed in a new 800 f5.6 lens to my 1DX III earlier this year I was setting up a charcoal grill to cook steaks when Fed Ex delivered the lens and the handy Kingsford charcoal bag and the grass around it made a practical target. The first image below was with no calibration offset and indicated slight front focusing, a +3 offset provided the best focus accuracy for this body and lens. The same lens on my 5DS R requires a +5.
Many years ago I used to calibrate Tektronix 7000 series oscilloscope mainframes which were the high end market leader because of their performance and versatile range of measurement applications through the very wide range of plug in modules. I had the required calibration standards, Tektronix standard value calibration plug ins, and the calibration fixtures which allowed calibrating every scope mainframe and plug in module to manufacturer standards which meant that any scope mainframe and any module could be paired and would meet STATED specifications over its measurement capability range. HOWEVER if a specific scope mainframe was calibrated with a specific set of plug-ins, an even higher standard of measurement accuracy could be achieved. That is very much like the way the Canon lenses and bodies are set up, the better lenses and bodies will work pretty darn well without specific calibration and will be sufficient for most purposes but to achieve the very best of which they are capable then it makes sense to spend a little time perfectly matching the lens and body as a system.
Again, for most lenses and situations it won't matter but if you are shooting with a very shallow depth of field and even more so when you need to use the full DoF that the lens provides at a certain focal length, aperture, and distance then calibration needs to be dead on.
Rodger
