Backfocus difference between Sigma and Minolta 50/1.4

[Jonathan K]

Dear David, dear forum members,

about half a year ago I started a thread about backfocus issues with the 50/1.4, and I was happy to get a great number of useful and informative answers.
Two days ago, I purchased the Sigma 50/1.4 HSM, which I tried out at Photokina.

I notices something weird: The backfocus issue is by far less pronounced with the Sigma than with my old Minolta (1st generation).
The Sigma is about spot on, as far as I can judge.

Question: If the backfocus issue is (as we found out in my last thread) supposed to be a camera issue, why do the two lenses behave differently?
I tried them out on my KM 7D, which was calibrated with the old Minolta 50/1. 4 (among others) at the German Minolta Service in Bremen and has shown no BF after that: Both lenses work well?

What do I get wrong?


Thanks for your ideas,

Jonathan

[David Kilpatrick]

The Sigma has minimal focus shift with stop down/open up. Remember, when you fit a 50mm f1.4 lens to your KM7D, the focusing and viewing (screen) are done at an effective f/7.1 (AF) and f/4.5 (the screen). You can never actually see the true focus of the lens at f/1.4 since the focusing screen can not show it, and it will more or less agree with the AF sensor anyway. The factory calibration done by KM on the 7D was for a 50mm f1.7 lens at f/7.1 and 2.9m distance, but if you sent them a 50mm f1.4 to match to the body, they may have adjusted the calibration so the lens is correctly focused when using a wider aperture like f/2 or f/2.8 (the bench test still uses f/7.1 because that's the widest aperture the 7D AF sensors can 'see').

I have both the Sigma and the 50mm f1.4 here now, and with the f2.8 centre sensor, there is no significant difference on the A900 at the 'safe' working apertures for the AF system (f2.8 to f8).

The Sigma is an exceptional lens but I have been surprised by the quality of my 50mm f1.4 Minolta when correctly focused by the A900. I suggest that using any body which lacks an f/2.8 centre AF sensor to check backfocus calibration may produce misleading results. The biggest advantages of the Sigma are its lack of vignetting (it still vignettes but the centre circle of bright illumination is larger), excellent central resolution wide open (the edges are never as good), excellent geometry (under -1% distortion at normal working distances), true f/1.4 (across the frame it's a good half stop brighter than the Minolta), and HSM (seems to work very well).

But - I can get a sharper picture corner to corner from the Minolta when stopped down.

David

[01af]

Question: If the backfocus issue is (as we found out in my last thread) supposed to be a camera issue, why do the two lenses behave differently?

When you're not aware of the problem of focus shift then any reasoning about back-focus issues is completely pointless. Which apertures, which ranges, and what kind of light are you talking about?

I received my A900 ten days ago, and currently I am in the midst of fine-adjusting the auto-focus for all my lenses. After ten days, I am far from finished ... and I have not even touched any of the zoom lenses yet. The focus is shifting with the colour of the light, the distance range, and the aperture ... and maybe also with the brightness level (not quite sure about this one yet). With regard to aperture-induced focus shift, the Minolta AF 20 mm 1:2.8 and the AF 50 mm 1:1.4 exhibit strong focus shift, the AF 24 mm 1.2.8 shows minor focus shift, and the AF Macro 50 mm 1:3.5 has virtually no focus shift at all. To name a few. Boy, the process of AF micro-adjusting really makes you aware of these issues!

With all lenses I have tested so far, the aperture-induced focus shift (if present) introduces more back focus at smaller apertures---i. e. the plane of focus will move towards the background when stopping down. On my A900 body, my AF 50 mm 1:1.4 requires an AF micro-adjustment value of -2 at full aperture but -7 at f/8; the focus shift is not even between f-stops, the strongest shift occurs between f/4 and f/5.6. So---when comparing the back focus of different lenses, make sure you are comparing them at the same apertures. And do the comparisons at various apertures, because the degree of back focus can be, for example, the same at full aperture but different at medium aperture ... or vice versa ... or whatever.

Or better yet---go out and shoot something. Because lens testing and focus adjustment quickly will become a bottomless pit ...

-- Olaf

[01af]

... I have been surprised by the quality of my 50 mm f1.4 Minolta when correctly focused by the A900.

Same here. In the past, I wasn't very happy with my AF 50 mm 1:1.4. I always preferred the AF 50 mm 1:1.7 because my Dynax 7D focuses this one more accurately than the former. From focus-bracketing tests, I was aware that my 50/1.4 is the better performer than my 50/1.7 (not true generally due to sample variation; 'my' being the operational words here)---but I never knew how fantastic the 50/1.4 really is!

This experience alone justifies the expense for the A900. I will never again buy a DSLR body without AF micro adjustment. The major point of the A900 is neither full-frame nor 24 MP but accurate auto-focusing.

-- Olaf

[Dr. Harout]

...This experience alone justifies the expense for the A900. I will never again buy a DLSR body without AF micro adjustment. The major point of the A900 is neither full-frame nor 24 MP but accurate auto-focusing.

-- Olaf

A very good point, Olaf. Thanks for emphasizing that.

[Jonathan K]

Dear David,

The Sigma has minimal focus shift with stop down/open up. Remember, when you fit a 50mm f1.4 lens to your KM7D, the focusing and viewing (screen) are done at an effective f/7.1 (AF) and f/4.5 (the screen). You can never actually see the true focus of the lens at f/1.4 since the focusing screen can not show it, and it will more or less agree with the AF sensor anyway. The factory calibration done by KM on the 7D was for a 50mm f1.7 lens at f/7.1 and 2.9m distance, but if you sent them a 50mm f1.4 to match to the body, they may have adjusted the calibration so the lens is correctly focused when using a wider aperture like f/2 or f/2.8 (the bench test still uses f/7.1 because that's the widest aperture the 7D AF sensors can 'see').

How do you find out in how far a lens is prone to exhibit focus shift? Just a matter of trying out?
I did not know that the AS sensors of the KM7D are sensitive only to f7.1 equivalent. I thought they were f5.6?
But you are right, there is no real point in comparing the lenses on the 7D. Fact is: the sigma is useable whereas the Minolta does not work in everyday life. Every single shooting with the Minolta 50/1.4 turned out to be a nightmare, even on the 7D after calibration, let alone on the alpha 700, which has not been calibrated yet.
So, after having read your observation regarding the amount of focal shift with the Sigma in comparison to the Minolta, wouldn't it be logical to consider this a main positive caracteristic of the lens, making it the "better", more reliable lens in every day life? At least that is what I notice when I shoot.

I have both the Sigma and the 50mm f1.4 here now, and with the f2.8 centre sensor, there is no significant difference on the A900 at the 'safe' working apertures for the AF system (f2.8 to f8).

The Sigma is an exceptional lens but I have been surprised by the quality of my 50mm f1.4 Minolta when correctly focused by the A900. I suggest that using any body which lacks an f/2.8 centre AF sensor to check backfocus calibration may produce misleading results. The biggest advantages of the Sigma are its lack of vignetting (it still vignettes but the centre circle of bright illumination is larger), excellent central resolution wide open (the edges are never as good), excellent geometry (under -1% distortion at normal working distances), true f/1.4 (across the frame it's a good half stop brighter than the Minolta), and HSM (seems to work very well).

But - I can get a sharper picture corner to corner from the Minolta when stopped down.

David

Pleased to hear about that. So I might not sell my Minolta lens yet. But, as you stated, the Sigma is a great lens to use. In the field it seems to work much better than the Minolta, at least for low light photography, which is one of the main reasons for me to own it.
One more thing:
I recently read that the focus adjustement on the A900 only works with Minolta/Sony lenses. Is that also true for the newest generation of 3rd party lenses, like this 50/1.4?

Thanks again for your explanations,

Jonathan

[Jonathan K]

I received my A900 ten days ago, and currently I am in the midst of fine-adjusting the auto-focus for all my lenses. After ten days, I am far from finished ... and I have not even touched any of the zoom lenses yet. The focus is shifting with the colour of the light, the distance range, and the aperture ... and maybe also with the brightness level (not quite sure about this one yet). With regard to aperture-induced focus shift, the Minolta AF 20 mm 1:2.8 and the AF 50 mm 1:1.4 exhibit strong focus shift, the AF 24 mm 1.2.8 shows minor focus shift, and the AF Macro 50 mm 1:3.5 has virtually no focus shift at all. To name a few. Boy, the process of AF micro-adjusting really makes you aware of these issues!

With all lenses I have tested so far, the aperture-induced focus shift (if present) introduces more back focus at smaller apertures---i. e. the plane of focus will move towards the background when stopping down. On my A900 body, my AF 50 mm 1:1.4 requires an AF micro-adjustment value of -2 at full aperture but -7 at f/8; the focus shift is not even between f-stops, the strongest shift occurs between f/4 and f/5.6. So---when comparing the back focus of different lenses, make sure you are comparing them at the same apertures. And do the comparisons at various apertures, because the degree of back focus can be, for example, the same at full aperture but different at medium aperture ... or vice versa ... or whatever.

Or better yet---go out and shoot something. Because lens testing and focus adjustment quickly will become a bottomless pit ...

-- Olaf

Hi Olaf,

Sounds a bit scary, if you ask me. Well, thanks to David here on the forum, I am pretty aware of focus shift and its consequences, and yes, obviously, I tried both lenses at the same aperture... Same light, same conditions etc, even on a tripod. But I do not aim for a absolute scientific understanding, I just wish to know my equipment and to know what I can expect from it.
So far, I have left the Minolta 50/1.4 at home, since it did not deliver useable pictures (or, to be fair, let's say: I was not able to get useable pictures using that lens), at least not in an acceptable degree, considering the conditions I was shooting. That was the main motivation to try out the Sigma and, after playing around with it, it seems to exhibit focal shift in a lesser extent, which is in line with David's judgement. It just works better in every day life, as I stated in a post before.

The main question remaining for me: Since I decided to send my Alpha 700 for calibration in the next 6 months (before the warranty runs out) after having experienced serious issues with 2 lenses, the 100mm Macro and the 50/1.4, how will the performance of this Sigma 50/1.4 change after the calibration? I have no idea about the 100mm Macro and its focal shift caracteristics, but at least it is a f2.8 and one should be able to rely on the AF module for this one. But: if the camera will be calibrated for the two Minolta lenses, could it be that the Sigma will fail to work?

A bit speculative, I know, but....

My next camera certainly will have micro adjustement, no question there. Absolutely valid point. Curious to see, what the speculated A800 (if it shows at PMA) will have in this respect?

Cheers Jonathan

[David Kilpatrick]

I can not speculate, but I think micro AF adjustment is likely in future higher end bodies. Also, I think that special attention to focus shift will be paid in all future lens designs. Sony read this stuff. I am writing about it in the British Journal, and I am the first UK journalist to do so as far as I know. I have written about other issues (notably the issue with focusing screens and d-o-f) and seen the results propagated throughout the net and throughout photo magazines worldwide. In every case, I am not responsible for identifying the problem - I'm a journalist, not a technician. But I have been able to explain things, express them clearly and demonstrate them, and get access to respected media to propagate the knowledge.

Eventually, this reaches Japan - they should already KNOW all this stuff! But it seems they only react when the rest of world becomes aware. Now they know that Joe Photographer is becoming aware of aperture related focus shift, and the value of micro AF adjustment, they are working on it. I think that in two or three years we will see AF adjustment capable of dealing with aperture and focus distance changes

David

[bakubo]

On my A900 body, my AF 50 mm 1:1.4 requires an AF micro-adjustment value of -2 at full aperture but -7 at f/8; the focus shift is not even between f-stops, the strongest shift occurs between f/4 and f/5.6.

I have an A700 so I don't know the details of how the A900 (or Nikon or Canon) AF micro-adjustment works. Are you saying that the A900 allows you to set multiple values for one lens depending on aperture, distance, color of light, etc.? That could turn out to be a whole bunch of combinations to test and store in the A900's memory. Does the A900 work this way?

[David Kilpatrick]

No, one setting only per lens. But you can memorise some settings, and use them manually. One day, there will be vectorised settings. Or lenses with no focus shift!

David

[01af]

On my A900 body, my AF 50 mm 1:1.4 requires an AF micro-adjustment value of -2 at full aperture but -7 at f/8 ...

Are you saying that the A900 allows you to set multiple values for one lens depending on aperture, distance, color of light, etc.?

No, unfortunately it's only one single adjustment value per lens (for up to 30 lenses). So if I'm changing the aperture from, say, f/1.4 to f/8 on the 50 mm lens then I have to go into the menu and change the AF adjustment value manually. It's a tedious procedure---but at least I can do something about AF accuracy. I have set up the menu so it will keep track where it was last time and immediately bring me to the AF micro-adjustment setting whenever I hit the Menu button.

That could turn out to be a whole bunch of combinations to test and store in the A900's memory.

That's right. However it gets stored in my head's memory, not in the A900's.

An ideal AF micro-adjustment system would provide a two-dimensional look-up table with about twelve or twenty values for every prime lens, and a three-dimensional look-up table for every zoom lens---one dimension for distance, one for aperture, and one for focal length. Furthermore the camera would have an automated calibration procedure so it can fill the table's column for one distance and one focal length automatically. The user would place the camera on a tripod at a certain distance before a brick wall or something, hit a button, and the camera would do the rest.

So far, I am testing lenses at medium distance only---about 100 times the focal length usually. But I know back-focus issues are worse at medium and long distances and less serious at close-up distances (that's why the so-called 'battery test' or 'cereal box test' on the kitchen table is so pointless). But I have not yet figured out if that also means different AF adjustment values. Maybe we can do without the distance dimension after all ...

-- Olaf

[bakubo]

No, unfortunately it's only one single adjustment value per lens (for up to 30 lenses). So if I'm changing the aperture from, say, f/1.4 to f/8 on the 50 mm lens then I have to go into the menu and change the AF adjustment value manually. It's a tedious procedure---but at least I can do something about AF accuracy. I have set up the menu so it will keep track where it was last time and immediately bring me to the AF micro-adjustment setting whenever I hit the Menu button.

For me that would be totally unusable. It seems that it would be faster to just use manual focus, wouldn't it?

That's right. However it gets stored in my head's memory, not in the A900's.

Whew, you must have a great memory? You remember the settings for several lenses at different aperture/distance combinations? Also, you mentioned that the color of the light also has an effect. So, that would mean a different setting for each combo of lens/focal length/aperture/distance/light color.

An ideal AF micro-adjustment system would provide a two-dimensional look-up table with about twelve or twenty values for every prime lens, and a three-dimensional look-up table for every zoom lens---one dimension for distance, one for aperture, and one for focal length. Furthermore the camera would have an automated calibration procedure so it can fill the table's column for one distance and one focal length automatically. The user would place the camera on a tripod at a certain distance before a brick wall or something, hit a button, and the camera would do the rest.

Yes, that would be cool!

So far, I am testing lenses at medium distance only---about 100 times the focal length usually. But I know back-focus issues are worse at medium and long distances and less serious at close-up distances (that's why the so-called 'battery test' or 'cereal box test' on the kitchen table is so pointless). But I have not yet figured out if that also means different AF adjustment values. Maybe we can do without the distance dimension after all ...

Yes, I have noticed that AF doesn't usually work so well at very long distances. It seems that it rarely works at infinity, for example. Often, if it is a newer lens that allows to focus past infinity, the camera will usually do that. Images at infinity are usually soft so I switch to manual focus. With the A700's vf though it is not always great for using mf.

[01af]

No, unfortunately it's only one single adjustment value per lens (for up to 30 lenses). So if I'm changing the aperture from, say, f/1.4 to f/8 on the 50 mm lens then I have to go into the menu and change the AF adjustment value manually. [...]

For me that would be totally unusable.

Well ... in the sunshine I normally use apertures between f/5.6 and f/11 but sure not f/1.4. When shooting sports in a barely lit indoor arena then I'm at f/1.4 or f/2 but never at f/5.6. So in most circumstances it's not as unusable as it may seem.

It seems that it would be faster to just use manual focus, wouldn't it?

Definitely not. It's neither faster nor more accurate than micro-adjusted auto-focus. And you cannot see aperture-related focus shift anyway when focusing manually at full aperture. Except when using the DOF pre-view button---but have you ever tried to focus manually in stopped-down mode ... accurately?

That's right. However it gets stored in my head's memory, not in the A900's.

Whew, you must have a great memory? You remember the settings for several lenses at different aperture/distance combinations?

No, not really. Of course I am writing down my AF micro-adjustment test results. When I'm done adjusting all my lenses then I'll make a nice card-board reference chart with all the values, to carry along with me in my photo bag. And maybe it' not going to be so much to memorize anyway.The camera stores one value per lens, and you only need to remember which lenses exhibit serious focus shift and to which degree. Simply subtracting about 4 or 5 from the full-aperture adjustment value seems to do the trick for medium and small apertures in most cases.

Also, you mentioned that the color of the light also has an effect.

Yes. But for now I am simply ignoring this parameter. I just make sure not to do any focus testing under tungsten or fluorescent light or on brightly coloured targets.

-- Olaf