Copy to copy variation

Understood, thanks for the explanations. Let me see if we can do a second run with properly calibrating using MFA and if that makes a difference.

FEChariot wrote in post #16144658
It's far better to shoot a target with contrast parallel to the sensor plane and run the tape at a 45 far outside the active AF point but still in the frame.

Great tip!

ElectronGuru wrote in post #16145188
Great tip!

I actually did not get his tip. Can you or him explain?

Actually, since live view does not use MFA in anyway and I have always used live view 10x manual focus for all of these tests, havent I already proven the lens is bad? I mean if I MFA the lens how will that help anything, I dont think I can accurately shoot a res chart with PDAF from view finder. Does that make sense?

Here are our findings after MFA:

- Using DOT tune method, bad lens yields +3 MFA, good lens yields -3 MFA.
- Using the view finder, shooting 5 brackets of MFA in each lens makes virtually no difference (i.e. if bad lens had a DOT tune result of +3, we tried 0,+1,+2,+3,+4,+5 etc). That being said, after the +10, the pictures get blurry.
- We used a focusing distance of 2.5 meters (MFD for the lens is 1.9meters).
- Now, here is the interesting part, when view finder is used, pictures from both lenses, with or without MFA, is virtually indistinguishable (even my extremely picky eyes cannot tell the difference). Granted, we only shot F2.0, not sure what happens with 2.8 or 4.0 (my friend had to go, we'll continue tonight).
- Using the extender, the DOT tune method does not work. That is, dialing -20 or +20 in MFA still causes the focus to beep. [B]For the good lens[B], if the original MFA of -3 were used, the pictures would get blurry. [B]For the bad lens[B], if the original MFA of +3 were used, the pictures would get sharp.
- Using the extender, with no MFA for the good lens and with +3 MFA for the bed lens, both lenses perform similar. In fact, the bed lens is a bit sharper.

Now I have no idea what the heck is going on, this has confused me a lot. I always thought that live view 10x was the way to do these tests, but I don't understand the discrepancy between view finder and live view, especially, in the extender case, I don't understand why the original MFA value for the bad lens improve the sharpness but that is not true for the good lens.

Any help is greatly appreciated.

FEChariot,

Thanks for bringing this up, he may end up keeping this lens after all if it turns out the lens is good.

TheLensGuy wrote in post #16146188
Here are our findings after MFA:

- Using DOT tune method, bad lens yields +3 MFA, good lens yields -3 MFA.
- Using the view finder, shooting 5 brackets of MFA in each lens makes virtually no difference (i.e. if bad lens had a DOT tune result of +3, we tried 0,+1,+2,+3,+4,+5 etc). That being said, after the +10, the pictures get blurry.
- We used a focusing distance of 2.5 meters (MFD for the lens is 1.9meters).
- Now, here is the interesting part, when view finder is used, pictures from both lenses, with or without MFA, is virtually indistinguishable (even my extremely picky eyes cannot tell the difference). Granted, we only shot F2.0, not sure what happens with 2.8 or 4.0 (my friend had to go, we'll continue tonight).
- Using the extender, the DOT tune method does not work. That is, dialing -20 or +20 in MFA still causes the focus to beep. [B]For the good lens[B], if the original MFA of -3 were used, the pictures would get blurry. [B]For the bad lens[B], if the original MFA of +3 were used, the pictures would get sharp.
- Using the extender, with no MFA for the good lens and with +3 MFA for the bed lens, both lenses perform similar. In fact, the bed lens is a bit sharper.

Now I have no idea what the heck is going on, this has confused me a lot. I always thought that live view 10x was the way to do these tests, but I don't understand the discrepancy between view finder and live view, especially, in the extender case, I don't understand why the original MFA value for the bad lens improve the sharpness but that is not true for the good lens.

Any help is greatly appreciated.

FEChariot,

Thanks for bringing this up, he may end up keeping this lens after all if it turns out the lens is good.

Live View 10x with manual focusing is the way to ensure that any difference is due to the lens' inherent sharpness, rather than due to its AF being slightly off. Ideally Live View (contrast) focusing should yield equivalent results because it also uses the output of the actual image sensor. Using the camera's phase-detection focusing uses separate AF sensors rather than the main images ensor, and therefore is dependent on the calibration of the lens and camera, and as a result is unreliable for this kind of comparison.

Might be best just to have him send his camera body and lens to Canon for calibration. If it has shown to be sharp at times, it could be something internal with the lens... Failing PCB or something.

I have been surprised when sending in bodies or lenses what Canon does with them... Even for free check/cleans through CPS. Last time my 5D3 went in they replaced the entire AF mirror box assembly. Prior, they replaced internal components and adjusted positioning of the CMOS and other stuff.

Lens wise, my 1.4x III had pins and PCB's replaced inside for whatever reason... Other adjustments to other lenses and stuff as well that were functioning perfectly.

Could be something like that inside the lens or some type of decentering causing focus to be off at times. Not sure.

Or he can just exchange it for another.

Just throwing out some random thoughts

TheLensGuy wrote in post #16145848
I actually did not get his tip. Can you or him explain?

Take a ruler and look at it through your viewfinder. Then place the center of the center AF point dead center on 12". You will notice that the bottom of that AF point (especially since the actual sensor is larger than the rectangle in the view finder) is first hitting a point closer than 12" and that is what your camera will focus on.

TheLensGuy wrote in post #16146188
Here are our findings after MFA:
snip

Here watch this video in all 3 parts. Its very informative:

http://www.youtube.com​/watch?v=iAx86nblZ2g

Not that when using manual focus with focus confirmation, you get the best result from the phase detection AF, which can be off compared to best focus on the actual image sensor - the focus confirmation doesn't know how far off the phase detection is. It only knows that it is "good focus" on the phase-detection sensor which is a totally different path for the light beams.

Using the phase detection for focus confirmation when doing manual focus is basically the same thing as auto-focus, but with you changing the focus instead of the camera sending instructions to the motor in the lens.

That is why the lens sharpness test should be done by getting best visual sharpness on the image sensor - which is where live view comes in.

archer1960 wrote in post #16146259
Live View 10x with manual focusing is the way to ensure that any difference is due to the lens' inherent sharpness, rather than due to its AF being slightly off. Ideally Live View (contrast) focusing should yield equivalent results because it also uses the output of the actual image sensor. Using the camera's phase-detection focusing uses separate AF sensors rather than the main images ensor, and therefore is dependent on the calibration of the lens and camera, and as a result is unreliable for this kind of comparison.

So then what you are saying is, if using live view lens A is sharper than B, it is not because of focus calibration.

No - focus calibration is just to get the phase detect AF to correctly aim at the surface of the image sensor.

Live view focus on the other hand does not involve any phase detect AF.

But live view focus can still show issues with the lens, requiring Canon to pick the lens to pieces and put it together again, in case an element or two inside the lens aren't properly aligned.

What I meant was, if lens A is sharper than lens B, using live view, it means lens B has some sort of an inherent problem, which cannot be helped with the MFA the body will do. Is that correct? I didn't mean factory calibration, I meant the manual MFA calibration we do.

Yes, if lens A is sharper with you set the focus visually using live view, then lens A is "better" when it comes to sharpness. Live view short-circuits any errors from the phase detection AF.

TheLensGuy wrote in post #16146712
So then what you are saying is, if using live view lens A is sharper than B, it is not because of focus calibration.

Yes. If you have gotten live view focus as good as it can be (and you need to do it multiple times, multiple shots, etc to be sure), and there is still a significant difference in the lens sharpness, you can be sure there is something else going on that is unrelated to AF calibration.