Test to see if lens needs AF MA on 7D

Roy C

Roy C

Messages
3,383
Edit My Images
No
Please note:-this is not suppose to be a method of performing MA but rather a quick test to see if MA is needed.

I was reading on another forum that a quick test to see if a lens needs MA is to set up a target (box with printed letters....) in good light and with a tripod (sensible distance and lowish ISO) take a few shots with normal AF (starting OOF of course) and then Switch to Live View and 10x magnification with LiveMode autofocus and take some shots. If both sets are equally sharp then Lens/Body is fine and do not need MA. If the live view shot are sharper then you could need MA. IS THERE ANY MERIT IN THIS METHOD ?
 
Personally I wouldn't bother go out and use it and only start investigating if you see issues in real world use, test shooting photo charts is just opening yourself up to a world of pain.

Others will disagree etc.
 
a1ex2001 said:
Personally I wouldn't bother go out and use it and only start investigating if you see issues in real world use, test shooting photo charts is just opening yourself up to a world of pain.

Others will disagree etc.
Click to expand...
I am not interested about weather anyone is bothered to carry out the test or not as that was not what I was asking, my question was "IS THERE ANY MERIT IN THIS METHOD ? " e.g.(comparing Phase Detection vs Contrast Detection Autofocus).
 
Yes there is merrit in it from what I've read on the subject it seems quite popular. (I still wouldn't do it though ;) )
 
a1ex2001 said:
Yes there is merrit in it from what I've read on the subject it seems quite popular.
Click to expand...
The reason I am asking is because I have often heard of comparing normal Phase AF against manual focussing in live view at 10x but never against live view AF.
 
Yes, there is merit in it. There are various ways, and none of them at all difficult, but you have to know what you're doing.

The key aspect is the target, which must be flat and square to the camera, shot at a sensible distance, ie not some angled or three dimensional target at close range. Then at least you know that when it's right, it is right, and not skewed by the test itself.
 
Roy C said:
The reason I am asking is because I have often heard of comparing normal Phase AF against manual focussing in live view at 10x but never against live view AF.
Click to expand...

It works because the phase detection is done by separate sensors that have to be the exact same distance from the end of the lens as the sensor plane.

The way I understand it, small discrepancies in these distances are what can cause the consistent front or back focussing that MA is used to counteract.

Since contrast detect uses the output from the sensor itself, there is no chance of accuracy problems like that.
 
HoppyUK said:
Yes, there is merit in it. There are various ways, and none of them at all difficult, but you have to know what you're doing.

The key aspect is the target, which must be flat and square to the camera, shot at a sensible distance, ie not some angled or three dimensional target at close range. Then at least you know that when it's right, it is right, and not skewed by the test itself.
Click to expand...

squishy said:
It works because the phase detection is done by separate sensors that have to be the exact same distance from the end of the lens as the sensor plane.

The way I understand it, small discrepancies in these distances are what can cause the consistent front or back focussing that MA is used to counteract.

Since contrast detect uses the output from the sensor itself, there is no chance of accuracy problems like that.
Click to expand...
Thanks Richard and Adam, you have explained it nicely :thumbs::thumbs:
 
Yes, it's a very good method for checking.

The real reason why the contrast-detect method used in LiveView is so much more accurate is that it involves feedback.

With the phase-detect focusing the camera figures out how far the lens needs to move and tells the lens to move that distance, then the lens moves as it's been ordered to. There is no subsequent check to see if the focus is OK.

With the contrast-detect method the camera tells the lens to move a bit, checks to see if the focus is OK, tells it to move some more, checks, repeats loop until there is no further improvement. Better accuracy, but dreadfully slow.
 
I did it the lazy way and just took the same shots but at -/+ MA and saved the best setting after viewing on my pc. Been perfect since i did it :)

edit; Didn't read your please note bit :bonk:

Yes that method was the one i used when first having problems with my 7D focus but ended up sending it back but it was nothing to do with MA. My focus was all over the place and the body needed calibrating.
Sorry for going a bit off subject Roy
 
Last edited:
hollis_f said:
Yes, it's a very good method for checking.

The real reason why the contrast-detect method used in LiveView is so much more accurate is that it involves feedback.

With the phase-detect focusing the camera figures out how far the lens needs to move and tells the lens to move that distance, then the lens moves as it's been ordered to. There is no subsequent check to see if the focus is OK.

With the contrast-detect method the camera tells the lens to move a bit, checks to see if the focus is OK, tells it to move some more, checks, repeats loop until there is no further improvement. Better accuracy, but dreadfully slow.
Click to expand...

Yes, that's true, but it's not the reason AF calibration is needed. You could always get the camera to double-check with phase-detect just by refocusing again, but if the calibration is out it will still be wrong.

The difficulty is that phase-detect uses uses a simulated focus plane, as Squishy said, whereas contrast-detect uses the actual focus plane, ie the sensor.

Phase-detect is very fast, and works in quite poor light, but it needs a lens that does exactly what it's told and obviously the calibration must be right. Lens accuracy can also vary according to focusing distance, hence the problem with abnormally close targets, and zooms can vary with focal length - you just have to choose the best compromise, but you have to be extremely critical for that to be much of an issue.

Contrast-detect automatically compensates for all these things, but is slow and needs good light. Both systems need a contrasty target to lock on to - neither will focus at all if pointed at a plan wall, for example.
 
hollis_f said:
With the phase-detect focusing the camera figures out how far the lens needs to move and tells the lens to move that distance, then the lens moves as it's been ordered to. There is no subsequent check to see if the focus is OK.
Click to expand...

That's something that is written frequently but isn't strictly correct Frank...there is some checking.

A test scenario for you that I've tried myself. I used a slow focussing lens (the EF180/3.5L Macro). Focus the lens close to MFD (on your hand for instance). Move your hand out of shot and initiate a "one shot AF move" on a target at infinity and then immediately block the path with your hand again....the lens will return to focus on your hand and not at the infinity position it had been commanded to.

Spooky eh?

It would appear that it will recheck when in a focus search operation but not if the original calculation gave a definitive result.

Bob
 
Last edited:
My main issue with MA was finding constancy throughout the range.
 
hollis_f said:
Well, you live and learn. I've always thought that the phase-detect AF system was 'open-loop'. And I was wrong.

Here's a great article explaining the differences.
Click to expand...

Good link Frank, cheers :)

I like the last bit where Nikon's on-sensor phase-detect patent is mentioned, that has now become reality in the new Nikon 1. And where he suggests we may be heading for more AF revolution than evolution. It's happening - exciting times for new cameras :thumbs:
 
You must log in or register to reply here.
Back
Top