Pages: [1] 2   Go Down

Author Topic: back/front focus and AF microadjustment  (Read 10980 times)

james_elliot

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 57
    • http://www.photo-lovers.org
back/front focus and AF microadjustment
« on: January 11, 2009, 11:00:27 am »

I have been evaluating all my lenses regarding front and back focusing, and I also tested the AF microadjustment of the 5D MarkII
The findings are quite surprising. It seems that non canon lenses can not be micro adjusted, and that front and back focusing depends on the focal length used.
The complete results are available here.
I would be very interested if anyone had done the same kind of tests, and what were the findings.
Logged

stever

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1245
back/front focus and AF microadjustment
« Reply #1 on: January 11, 2009, 12:09:21 pm »

thanks, this is just the kind of information i've been looking for

it looks like i've been wasting some time with resolution tests without checking front/back focus, particularly on my 40D that was suposedly "corrected" by Canon under warranty

my problem has been determining whether the problem is with body or lens, so i ordered a 50D.  it would be very interesting to know how much the "0" point for adjustment varies from camera to camera

subjectively, my zoom lens results are disappointing with the exception of the 24-105 on my 5D (but the 24-105 is disappointing on the 40D)

with some lenses there may a focus shift with aperture as well

it's not surprising that micro adjust only works on Canon lenses as only Canon lenses are identified in the EXIF data which is what they must be using to determine adjustment

my sense of this is unfortunately a little like the camera manufacturers ignoring the problem of dirty sensors until complaints reached an unbearable level before acknowledging the problem and finally doing something about it.  i hope this is the beginning of the realization that lenses for high resolution cameras must be improved, and not just a band-aid allowing looser focus tolerances

i'll report as soon as i get the 50D next week and do some testing and hope that others will contribute their results
Logged

rod edwards

  • Newbie
  • *
  • Offline Offline
  • Posts: 10
    • http://www.rodedwards.com
back/front focus and AF microadjustment
« Reply #2 on: January 11, 2009, 01:06:52 pm »

Hi !

Have just read your post about microfocus adjustment ...

You may also find my new post of interest.

[a href=\'index.php?act=findpost&pid=251036\']AF Microfocus Adjustment with Canon 1Ds Mark 3[/a]

My findings are that at minimum focus distance (close up), a camera / lens combination may need large amounts of micro adjustments (up to +/- 20) and at infinity the same correctly calibrated camera / lens combo will need little or no micro adjustment (+/- 0) ... This means focus distance (subject to camera) is very relevant to AF microfocus adjustment settings and therefore the current af micro adjustment settings (as of Jan 2009) have limited use without Canon also taking distance info into consideration.

Regards,

Rod  
« Last Edit: January 11, 2009, 01:19:21 pm by rod edwards »
Logged
Professional photographer specialisng in

JDClements

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 312
    • http://www.jdanielclements.com
back/front focus and AF microadjustment
« Reply #3 on: January 11, 2009, 06:31:52 pm »

This is very interesting, but isn't the back/front focusing problem due to loose tolerances in the distance from the front surface of the lens mount to the front surface of the sensor? In other words, an issue with the construction of the camera body?
Logged

rod edwards

  • Newbie
  • *
  • Offline Offline
  • Posts: 10
    • http://www.rodedwards.com
back/front focus and AF microadjustment
« Reply #4 on: January 12, 2009, 07:51:07 am »

Whatever is causing it - it needs fixing !  

Any idea if focus distance info can be read by Canon lens / body ? I think Nikon D lenses can interpret this info for flash function at least ... maybe Canon can too ?! If so, maybe Canon can plot a focus distance curve or similar ?

Here are the images from my own tests (50mb zip folder) with Canon 1Ds 3 (factory calibrated twice) plus 85mm f1.2 II lens (factory calibrated twice) using Lens Align Pro at 0.95m, 1.5m and 2.5m - manual focus, +/-0 af microadjustment and the needed +20 setting to achieve perfect close up focus :

 http://rodedwards.co.uk/samples/aftests.zip.

My other lenses also exhibit similar characteristics.

It appears that AF Microadjustment changes considerably with focus distance as well as focal length.  

Would love to know other people's thoughts ?  

Quote from: JDClements
This is very interesting, but isn't the back/front focusing problem due to loose tolerances in the distance from the front surface of the lens mount to the front surface of the sensor? In other words, an issue with the construction of the camera body?
« Last Edit: January 12, 2009, 07:56:31 am by rod edwards »
Logged
Professional photographer specialisng in

Paul Roark

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 398
back/front focus and AF microadjustment
« Reply #5 on: January 12, 2009, 05:25:42 pm »

Is there a good explanation -- best from Canon -- what this adjustment is compensating for?  I'm hearing that it makes a difference what lens is used, the distance from the target, and what focal length a zoom is at.  When I was testing older lenses on medium format bodies, the amount of focus shift as the lens was stopped down was amazing with some lenses.  On the other hand, if an f/1.2 lens needs adjustment to be in focus wide open, where is the focus shift coming from?  I'm guessing some of the variables I'm hearing are more likely caused by something that is more unique to how AF systems work.

Paul
www.PaulRoark.com
Logged

fike

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1391
  • Hiker Photographer
    • trailpixie.net
back/front focus and AF microadjustment
« Reply #6 on: January 12, 2009, 05:49:53 pm »

AF Micro Adjustment is a great feature.  We didn't know how much we needed until we got it.
 

It IS a great feature, but I don't think there is a lens problem as more than one poster has suggested.  These are good lenses and in most applications, the differences we are talking about won't cover the distance from someone's nose to their eyes.  SERIOUSLY! Now that we have such great pixel-peeping technology available to all of us, we use it and then think that we have been sold a poor quality lens or body. No, actually we have been sold a product that is so good that it reveals the limitations in other parts of the system.  I got excellent images with my 30D.  Now I can do even better with my 50D using AF Micro Adjust.  The lenses I used with the 30D and now on the 50D didn't suddenly become "bad copies."  There have always been variances in manufacturing, and for some very fine work, that may be a problem, but for the majority of our landscape work, we are seriously splitting hairs here.

There certainly is a good case for wanting better lens technologies, but I am afraid that the lens wars will not be as quick and easy as the megapixel wars--Moore's Law doesn't apply to optics.  

Instead, how about 32 Bit per channel color?  Anyone? Anyone?

<sarcastic rant is over>
Logged
Fike, Trailpixie, or Marc Shaffer

rod edwards

  • Newbie
  • *
  • Offline Offline
  • Posts: 10
    • http://www.rodedwards.com
back/front focus and AF microadjustment
« Reply #7 on: January 12, 2009, 06:21:48 pm »

Unfortunately, I don't just shoot landscapes - i shoot people as well. This is why i bought the 85mm f1.2 II high quality portrait lens - for portraits at varying distances with nice shallow focus.
However, it seems that my camera and lens (after twice being calibrated by Canon Pro UK) is unable to focus perfectly even after microadjustments. I also found this with other lenses and my new 5D2.
Whilst i realise I do expect a great deal from our equipment, I am a professional with demanding clients. I bought the pro grade Canon L lenses with a pro grade 1Ds 3 body not a Practika.
Surely expecting my images to 'mostly' be in focus is not expecting too much after spending 20,000 on Canon equipment in the last twelve months ?  
I think this is an issue and Canon need to sort it !  



Quote from: fike
AF Micro Adjustment is a great feature.  We didn't know how much we needed until we got it.
 

It IS a great feature, but I don't think there is a lens problem as more than one poster has suggested.  These are good lenses and in most applications, the differences we are talking about won't cover the distance from someone's nose to their eyes.  SERIOUSLY! Now that we have such great pixel-peeping technology available to all of us, we use it and then think that we have been sold a poor quality lens or body. No, actually we have been sold a product that is so good that it reveals the limitations in other parts of the system.  I got excellent images with my 30D.  Now I can do even better with my 50D using AF Micro Adjust.  The lenses I used with the 30D and now on the 50D didn't suddenly become "bad copies."  There have always been variances in manufacturing, and for some very fine work, that may be a problem, but for the majority of our landscape work, we are seriously splitting hairs here.

There certainly is a good case for wanting better lens technologies, but I am afraid that the lens wars will not be as quick and easy as the megapixel wars--Moore's Law doesn't apply to optics.  

Instead, how about 32 Bit per channel color?  Anyone? Anyone?

<sarcastic rant is over>
Logged
Professional photographer specialisng in

JDClements

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 312
    • http://www.jdanielclements.com
back/front focus and AF microadjustment
« Reply #8 on: January 12, 2009, 07:51:00 pm »

This is what I understand (which could be wrong): The AF system "looks" through the lens, uses its magic to decide exactly how far away it is focusing, then adjusts the lens accordingly. When it does this, it is programmed to expect the sensor to be in a very specific position. If the sensor is not where it is expected to be, you have front- or back-focusing. (If it is a little too far forward in the camera, it will back-focus, if it is a little too far back in the camera, it will front focus.)
Logged

Mark F

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 365
back/front focus and AF microadjustment
« Reply #9 on: January 12, 2009, 08:31:33 pm »

So this means that there is no point in doing the microfocus because if you fix one end of the focus scale you make the other end worse?

A few weeks back I posted a similar question as I was having trouble getting my 400mm to focus accurately close-up altough it does a good job at distance. At least I now know that I'm not the only one who has encountered this.


Quote from: rod edwards
.......
My findings are that at minimum focus distance (close up), a camera / lens combination may need large amounts of micro adjustments (up to +/- 20) and at infinity the same correctly calibrated camera / lens combo will need little or no micro adjustment (+/- 0) ... This means focus distance (subject to camera) is very relevant to AF microfocus adjustment settings and therefore the current af micro adjustment settings (as of Jan 2009) have limited use without Canon also taking distance info into consideration.

Regards,

Rod  
Logged
Mark

rod edwards

  • Newbie
  • *
  • Offline Offline
  • Posts: 10
    • http://www.rodedwards.com
back/front focus and AF microadjustment
« Reply #10 on: January 13, 2009, 03:46:55 am »

In reply to your your question Mark F... i agree, there seems little or no point trying to adjust af microfocus as when one end is in perfect focus, this other end goes soft. With long lenses, wide apertures and close focus distances, this is a REAL problem.
My af system (camera & lenses) were calibrated by Canon Pro UK Service centre (twice) so i presume this is a focus issue rather than faulty kit.
The only solution is to manual focus with Live View or maybe try a better focus screen like the Canon EC-L for the 1Ds 3.
I've tried this EC-L screen on my system and it is easier to focus, but you are then susceptible to user error and it's slower.
This seems to be a real issue that needs to be corrected by Canon urgently. The only way to do this is for us ALL to SHOUT very loudly to Canon to get this fixed.  

Quote from: Mark F
So this means that there is no point in doing the microfocus because if you fix one end of the focus scale you make the other end worse?

A few weeks back I posted a similar question as I was having trouble getting my 400mm to focus accurately close-up altough it does a good job at distance. At least I now know that I'm not the only one who has encountered this.
Logged
Professional photographer specialisng in

fike

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1391
  • Hiker Photographer
    • trailpixie.net
back/front focus and AF microadjustment
« Reply #11 on: January 13, 2009, 12:31:26 pm »

Quote from: Mark F
So this means that there is no point in doing the microfocus because if you fix one end of the focus scale you make the other end worse?

A few weeks back I posted a similar question as I was having trouble getting my 400mm to focus accurately close-up altough it does a good job at distance. At least I now know that I'm not the only one who has encountered this.

That's pretty interesting.  I adjusted my 100-400 a few weeks ago and from a purely subjective point of view, I think I have improved focus accuracy.  I used a very close subject for the calibration though--about 7 feet.  I'll need to see if this holds true out closer to infinity.  I think it is much harder to assess focus at longer ranges.  Visual acuity is affected by glare, atmosphere, etc..., so clearly assessing sharpness would be very hard.  Also, all the lens reviews and resolution charts are shot at more typical indoor ranges, where the environment can be controlled better.  If this is true, this entire method of analyzing telephoto zoom lenses would be demonstrated to be, at least partially, invalid.  

I am pretty curious to see where this discussion goes.
Logged
Fike, Trailpixie, or Marc Shaffer

AndyF

  • Newbie
  • *
  • Offline Offline
  • Posts: 44
    • http://
back/front focus and AF microadjustment
« Reply #12 on: January 13, 2009, 01:10:10 pm »

Quote from: JDClements
This is very interesting, but isn't the back/front focusing problem due to loose tolerances in the distance from the front surface of the lens mount to the front surface of the sensor? In other words, an issue with the construction of the camera body?
I'd (obviously) agree a small focus adjustment should be the needed, and it's to null out the difference in the distance to the focus sensor vs. the image sensor.  If that is the case, it should be a specific value for that camera body.  The tests people are reporting usually only show one A:B comparison.  There will be some small variation in focusing error every time the lens is re-focused and this variation might be mis-interpreted as the microfocus adjustment being "wrong".

The focus error might also depend on how far off-focus the lens was to start with, and i which direction the lens was focusing.  A single shot, compared to a single shot with a different lens wouldn't really be testing just the microfocus difference.

If someone is going to be testing their lenses in more detail, it would be worth checking the same body/lens in five scenarios: start with the lens focused fully short, fully long, slightly short, slightly long, and the final test: remove and remount the lens each time.  Run those five scenarios three times.  If the microfocus adjustment is the only correction needed, the errors will be the same in each of those setups.  But I don't think it will be.  There should also be some non-repeatability in the lens focusing.

Maybe the microfocusing adjustment should be the average of several different experiments, and the other errors need some other correction method.
Andy


Logged

rod edwards

  • Newbie
  • *
  • Offline Offline
  • Posts: 10
    • http://www.rodedwards.com
back/front focus and AF microadjustment
« Reply #13 on: January 13, 2009, 01:55:26 pm »

All very valid points Andy and I agree, to do any comprehensive test, all the variables need to be eliminated.
For my own personal lens tests with the 85mm lens (the above samples), i did indeed average the autofocus results over 5-10 images having first set the lens to infinity so it whirled back to close focus each time.
To be honest, if a lens seems to have the need for +20 microadjustment in any direction as my (apparently calibrated) lens does, i think it may be problematic at other distances after microadjustment.
I've spoken with Canon again today and they have been very helpful and say there is obviously something wrong with my lens / camera. I've just been doing some quick tests with my 5D2 and this needs around +7 adjustment with the same 85mm lens (not +20 like my 1Ds 3).
I'm taking them both in to Canon tomorrow for a thorough checking over. I also noticed different results on the 1ds3 using One Shot mode or Servo af ... but i guess that's another tin of worms all together  


Quote from: AndyF
I'd (obviously) agree a small focus adjustment should be the needed, and it's to null out the difference in the distance to the focus sensor vs. the image sensor.  If that is the case, it should be a specific value for that camera body.  The tests people are reporting usually only show one A:B comparison.  There will be some small variation in focusing error every time the lens is re-focused and this variation might be mis-interpreted as the microfocus adjustment being "wrong".

The focus error might also depend on how far off-focus the lens was to start with, and i which direction the lens was focusing.  A single shot, compared to a single shot with a different lens wouldn't really be testing just the microfocus difference.

If someone is going to be testing their lenses in more detail, it would be worth checking the same body/lens in five scenarios: start with the lens focused fully short, fully long, slightly short, slightly long, and the final test: remove and remount the lens each time.  Run those five scenarios three times.  If the microfocus adjustment is the only correction needed, the errors will be the same in each of those setups.  But I don't think it will be.  There should also be some non-repeatability in the lens focusing.

Maybe the microfocusing adjustment should be the average of several different experiments, and the other errors need some other correction method.
Andy
Logged
Professional photographer specialisng in

stever

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1245
back/front focus and AF microadjustment
« Reply #14 on: January 13, 2009, 11:05:00 pm »

i've just been reading the SLRgear article on focus and lens testing procedure - the lack of autofucus repeatability is disappointing and the testing procedure which levels the playing field for optics is a step further removed from what anybody can expect from a lens/camera combination in the real world - no wonder i can't duplicate the resolution test results

i wouldn't be unhappy if the next "feature" is focus bracketing (i've wanted it since i first tried Helicon Focus anyhow) - just automate the microfocus adjust like exposure bracketing

after sandbagging the tripod and locking up the mirror locking up the mirror to get critical focus, then finding that you can't rely on repeatable autofocus (even if improved somewhat by "microadjustment") and that the manual focus is too sensitive to be useful even if you're using liveview, it causes me to wonder if the manufacturers have quite got their priorities right
Logged

Mark F

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 365
back/front focus and AF microadjustment
« Reply #15 on: January 13, 2009, 11:09:07 pm »

I think that other factors come into play also - shutter speed, mirror lockup, loose lens collar,  and whether your camera is locked down tightly on your tripod. All of these can effect apparent sharpness so should be eliminated from the test.  At slower shutter speeds (1/30 or slower) I always use mirror lockup on my tripod or I lose sharpness. But that is not what I was referring to in my other post. Using mirror lockup and with my camera firmly locked on my tripod I still get soft images close up but sharp images at distance.
Logged
Mark

AndyF

  • Newbie
  • *
  • Offline Offline
  • Posts: 44
    • http://
back/front focus and AF microadjustment
« Reply #16 on: January 13, 2009, 11:44:23 pm »

Quote from: rod edwards
All very valid points Andy and I agree, to do any comprehensive test, all the variables need to be eliminated.
For my own personal lens tests with the 85mm lens (the above samples), i did indeed average the autofocus results over 5-10 images having first set the lens to infinity so it whirled back to close focus each time.
...
Your tests were pretty thorough then!
Two ways microfocus adjustment could be implemented, are (1) fine adjust the position where the focus sensors are.  This is certainly not what is happening in the camera but should result in perfect nulling over the entire zoom range.  (2) focus until the sensors indicate the image is in focus, then add or subtract a calibrated offset.  The focal point of a zoom lens may shift slightly through it's zoom range.  The calibrated microfocus offset will then be valid for that lens, at the zoom length it was measured at, but for any other zoom point it's an extrapolation and may not be exactly what that lens needs.  If indeed the value is scaled at all to match the lens setting.

It might therefore be possible for Canon to improve this with a software update.  Instead of storing one correction value for a lens on that body, store several that had been measured over the entire zoom range and from that set of values, perform a more accurate interpolation for the value needed at any particular setting.

Andy
Logged

rod edwards

  • Newbie
  • *
  • Offline Offline
  • Posts: 10
    • http://www.rodedwards.com
back/front focus and AF microadjustment
« Reply #17 on: January 14, 2009, 03:51:30 am »

I wonder if Canon lenses can read focus distance info ? I know Nikon D lenses do some sort of distance calculations for their flash systems - maybe Canon also use this type of info ? Anyone know ?

If so, i think it may be possible for the user to calculate the correct af microadjustment needed at minimum focus distance and then the camera could focus according to a 'corrective focus curve' or similar ? This curve could theoretically be able to be set by Canon by testing zeroed bodies and lens combinations and applying corrections accordingly via firmware. It may not be 100% perfect but would be much, much better for closer distances.

Nice idea eh ?!  But can Canon lenses / bodies read focus distance info ?  

Quote from: AndyF
Your tests were pretty thorough then!
Two ways microfocus adjustment could be implemented, are (1) fine adjust the position where the focus sensors are.  This is certainly not what is happening in the camera but should result in perfect nulling over the entire zoom range.  (2) focus until the sensors indicate the image is in focus, then add or subtract a calibrated offset.  The focal point of a zoom lens may shift slightly through it's zoom range.  The calibrated microfocus offset will then be valid for that lens, at the zoom length it was measured at, but for any other zoom point it's an extrapolation and may not be exactly what that lens needs.  If indeed the value is scaled at all to match the lens setting.

It might therefore be possible for Canon to improve this with a software update.  Instead of storing one correction value for a lens on that body, store several that had been measured over the entire zoom range and from that set of values, perform a more accurate interpolation for the value needed at any particular setting.

Andy
« Last Edit: January 14, 2009, 03:52:13 am by rod edwards »
Logged
Professional photographer specialisng in

AndyF

  • Newbie
  • *
  • Offline Offline
  • Posts: 44
    • http://
back/front focus and AF microadjustment
« Reply #18 on: January 14, 2009, 12:51:36 pm »

Quote from: rod edwards
...
Nice idea eh ?!  But can Canon lenses / bodies read focus distance info ?
...
Yes.  There is a product I heard of from Birger, a standalone Canon lens mount (when a lens is needed for some other kind of equipment).  Their software has PositionOfFocus and FocusDistance commands.

Lens mount:
http://www.birger.com/Merchant2/merchant.m...reen=ef232_home
http://www.birger.com/Merchant2/merchant.m...=ef232_softintf

So the camera can find out the focus distance.  However, knowing that, we can't actually do much until either Canon adds multi-point focus correction or the CHDK developers (http://chdk.setepontos.com/) succeed in porting features to the DSLR operating systems.  They've done this on the digicams and added scripts, focus bracketing, motion detection, etc.  They may have less success with the DSLRs though.

Andy
Logged

james_elliot

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 57
    • http://www.photo-lovers.org
back/front focus and AF microadjustment
« Reply #19 on: January 15, 2009, 12:19:39 pm »

Yes they can, and the information was available in the former Canon raw file format (CRW) that was used for example by the good old 300D. It is mandatory for the body to get the distance from the lens for E-TTLII to work.
However Canon doesn't seem to store the information anymore in its CR2 files. A real pity, even more so if you are using a software like DxO which needs this information to correct lens distorsion...

Quote from: rod edwards
I wonder if Canon lenses can read focus distance info ? I know Nikon D lenses do some sort of distance calculations for their flash systems - maybe Canon also use this type of info ? Anyone know ?

If so, i think it may be possible for the user to calculate the correct af microadjustment needed at minimum focus distance and then the camera could focus according to a 'corrective focus curve' or similar ? This curve could theoretically be able to be set by Canon by testing zeroed bodies and lens combinations and applying corrections accordingly via firmware. It may not be 100% perfect but would be much, much better for closer distances.

Nice idea eh ?!  But can Canon lenses / bodies read focus distance info ?
Logged
Pages: [1] 2   Go Up