SDQ Viewfinder and Live View

[dr.noise]

I did not fully understand if the "focus shift" is a doubt or a clearly observed optical property of some lenses.

I have to clear what I wrote. By focus shift I mean rotating the focus ring, i.e. the shift that occurs naturally.

A focus shift that occurs when the aperture value is changed is also real. I have a lens that clearly shows it, and the shift is so big that I really have to re-focus considerably to get acceptable result.

 

[rf-design]

Before the lens where possible designed for minimum focus shift. Now the picture change.

True, with the newer fast lens group moving electromechnalical solutions they could change the focus at exposure time. I have doubts that they doing this beside it is possible.

However, this does not mean that the focusing algorithm sets the lens at perfect focus at maximum aperture. Likely it sets the lens at some small offset so when the aperture closes down focus is perfect (as estimated by the USB dock data, maybe?). Meanwhile, the user does not see the small difference in the viewfinder. No need to move focus groups fast.

You are right, shifting focus with the already known target apture before is more clever solution.

Pixel counts have gone very high, making visible even the smallest deviations from peak lens performance. If a Sigma electronic lens is designed for minimum focus shift, there is no way to know if there has actually been a shift until after the sd Quattro shot is taken.

So the above solution could be used also by the SDQ because the lens and also its possible optimum focus is known in advance. This information could also be applied for MF mode. So in the best case a "focus shift" lens is corrected without knowing about.

DOF preview will come sometime. It is such a missing peace.

Basically I don't even bother to use anything on my sd Quattro except my only global vision lens, a 24-105 F4 Art. Works great. Someday, I'll get a manual lens mounted...

--
Tom Schum
Every day a new image.

 

[rf-design]

I did not fully understand if the "focus shift" is a doubt or a clearly observed optical property of some lenses.

I have to clear what I wrote. By focus shift I mean rotating the focus ring, i.e. the shift that occurs naturally.

A focus shift that occurs when the aperture value is changed is also real. I have a lens that clearly shows it, and the shift is so big that I really have to re-focus considerably to get acceptable result.

I think this is an underestimated lens issue to get worked with. Which lens show "focus shift"?

 

[dr.noise]

focus shift that occurs when the aperture value is changed is also real. I have a lens that clearly shows it, and the shift is so big that I really have to re-focus considerably to get acceptable result.

I think this is an underestimated lens issue to get worked with. Which lens show "focus shift"?

That is an old M42 lens - called Formula 5, 28 mm f/2.8

Since I have many M42 lenses, and have repaired or disassembled many of them, I can say I'm pretty sure there is no mechanical issue in this lens.

 

[Tom Schum]

focus shift that occurs when the aperture value is changed is also real. I have a lens that clearly shows it, and the shift is so big that I really have to re-focus considerably to get acceptable result.

I think this is an underestimated lens issue to get worked with. Which lens show "focus shift"?

That is an old M42 lens - called Formula 5, 28 mm f/2.8

Since I have many M42 lenses, and have repaired or disassembled many of them, I can say I'm pretty sure there is no mechanical issue in this lens.

I can't report on any Sigma lenses because they are all electronic, and the upcoming Cine line cannot be obtained in SA mount.

I suppose an owner of a higher end Canon or Nikon could research this with a Sigma lens because one or the other of these cameras actually sets the lens to the shooting aperture in manual mode:

1. Set lens to F5.6, manually focus and shoot.

2. Without changing focus, set aperture to F3.5 and shoot.

3. Compare the shots for absolute clarity.

Actually, I could do something like this with my sd Quattro in manual mode:

1. Set Sigma electronic lens to F5.6, manually focus and shoot.

2. Change aperture to F3.5, do not change manual focus, and shoot.

3. Compare the shots for absolute clarity. Any variation is focus shift vs aperture.

This sounds like a lot of work, and I've already tired myself out just writing this post.

 

[xpatUSA]

Actually, I could do something like this with my sd Quattro in manual mode:

1. Set Sigma electronic lens to F5.6, manually focus and shoot.

2. Change aperture to F3.5, do not change manual focus, and shoot.

3. Compare the shots for absolute clarity. Any variation is focus shift vs aperture.

That could be a non sequitur, Tom, depending on what you mean by "absolute clarity"?

By which I mean that your choice of f/5.6 is the MTF "sweet spot" for many lenses and the MTF wide-open is often noticeably much less.

Pardon my pedantry.

Just to exaggerate, suppose I did the same test on my 50mm f/2.8 Macro EX DG at f/5.6 then f/45 [sic]? Would the variation in clarity be due to focus shift vs aperture or would it be due to diffraction? Sounds like obfuscation I know - but still, going the other way from the sweet spot to wide-open brings the question of aberration into play, I would think.

This sounds like a lot of work, and I've already tired myself out just writing this post.

Yep, I know the feeling. ;-)

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[Tom Schum]

Actually, I could do something like this with my sd Quattro in manual mode:

1. Set Sigma electronic lens to F5.6, manually focus and shoot.

2. Change aperture to F3.5, do not change manual focus, and shoot.

3. Compare the shots for absolute clarity. Any variation is focus shift vs aperture.

That could be a non sequitur, Tom, depending on what you mean by "absolute clarity"?

By which I mean that your choice of f/5.6 is the MTF "sweet spot" for many lenses and the MTF wide-open is often noticeably much less.

Pardon my pedantry.

Just to exaggerate, suppose I did the same test on my 50mm f/2.8 Macro EX DG at f/5.6 then f/45 [sic]? Would the variation in clarity be due to focus shift vs aperture or would it be due to diffraction? Sounds like obfuscation I know - but still, going the other way from the sweet spot to wide-open brings the question of aberration into play, I would think.

This sounds like a lot of work, and I've already tired myself out just writing this post.

Yep, I know the feeling. ;-)

 

[xpatUSA]

Actually, I could do something like this with my sd Quattro in manual mode:

1. Set Sigma electronic lens to F5.6, manually focus and shoot.

2. Change aperture to F3.5, do not change manual focus, and shoot.

3. Compare the shots for absolute clarity. Any variation is focus shift vs aperture.

That could be a non sequitur, Tom, depending on what you mean by "absolute clarity"?

By which I mean that your choice of f/5.6 is the MTF "sweet spot" for many lenses and the MTF wide-open is often noticeably much less.

Pardon my pedantry.

Just to exaggerate, suppose I did the same test on my 50mm f/2.8 Macro EX DG at f/5.6 then f/45 [sic]? Would the variation in clarity be due to focus shift vs aperture or would it be due to diffraction? Sounds like obfuscation I know - but still, going the other way from the sweet spot to wide-open brings the question of aberration into play, I would think.

This sounds like a lot of work, and I've already tired myself out just writing this post.

Yep, I know the feeling. ;-)

 

[Meuh]

Yes I have it set up for back button focus. The rear AF button does the focus then the shutter only does the exposure release. Very handy for most subjects.

Excellent, thank you. That is good news. BBF is very handy, in particular on a camera with slow focus. The DP Merrills apparently don't support this feature, unfortunately.

No but it was super quick to turn af off and on

 

[rf-design]

I have read now some search hits and found that there are some discrepancies and a need for further understanding of the AF method and focus shift relation. This link:

https://photographylife.com/what-is-focus-shift

discuss the relation of the AF method and focus shift. But to me there is one specific misunderstanding. The classic AF module use only a central portion of the lens up to equivalent F4-5.6, and up to F2.8 for Canon for instance. So the focus shift of the lens does not disturb the accuracy of the AF.

"Why is this all important and what does phase detect AF have to do with the focus shift? As shown in the illustration above, focus shifts to the right (or away from the camera) when aperture is changed. Since phase detect AF leaves the lens aperture wide open while autofocusing, the sharpest focus plane is going to shift when you take a picture with the lens stopped down. This is potentially a big problem, because it means that you might not be able to achieve correct focus, unless you stop down enough to compensate the shift (by increasing depth of field)"

...since phase detect AF leaves the lens aperture wide open while autofocusing, the sharpest focus plane is going to shift when you take a picture with the lens stopped down.

So the AF focus result is possible not the best at wide open but there the resolution of the lens must be anyway less than apture closed to the value where the AF module stop working. So the F4-5.6 limited AF module does not get disturbed by focus shift.

For CDAF the situation depend on at what apture the AF operate. Because for CDAF the focus shift will direct impact the focus result if it is not done at the target apture. That could be one reason why Sony make both focus methods, PDAF and CDAF, at target apture.

PDAF on sensor use either a half pixel shadow, dual half pixel or angeled microlens for PD. But in difference to classic separate PDAF modules they use the full apture. So if SDQ is using full open apture they must correct for focus shift and will be incompatible in some manner for manual lenses with significant focus shift.

 

[D Cox]

PDAF on sensor use either a half pixel shadow, dual half pixel or angeled microlens for PD. But in difference to classic separate PDAF modules they use the full apture. So if SDQ is using full open apture they must correct for focus shift and will be incompatible in some manner for manual lenses with significant focus shift.

So don't use lenses with this fault.

I would guess that these are mostly older lenses with very wide maximum apertures -- such as a 50mm f/1.2 or similar. They would have given focus problems on a traditional SLR, too.

 

[Tom Schum]

I have read now some search hits and found that there are some discrepancies and a need for further understanding of the AF method and focus shift relation. This link:

https://photographylife.com/what-is-focus-shift

discuss the relation of the AF method and focus shift.

Thanks for the link, the article was very informative.

Also from the article:

"Some entry-level DSLRs do not provide the ability to change aperture in Live View mode, which would again result in focus shift when changing lens aperture."

In other words, the sd Quattro is entry-level because it lacks the capability to set the lens to shooting aperture while shooting. I can agree wholeheartedly with this!

Their list of workarounds is basically the same as what I have been doing. Notable to me is that the first suggestion is to shoot at max aperture, which I have found to work, and the last suggestion to use a manual lens:

1. Shoot at maximum aperture.

2. AF fine tune optimal aperture. (take the USB dock with you to the shoot and spend hours tuning up the lens to get best focus at the aperture you want to shoot at. Not to mention the time spent in SPP to examine the images...)

3. Use a slower lens. (Sigma's 50mm macro might be slow enough, so if I can show mine doing focus shift this recommendation might not hold.)

4. Use contrast detect AF. (Still not applicable to the sd Quattro because it does not allow the user to set the lens to the shooting aperture and then focus. Besides, can we set the camera to use only contrast detect AF? Anyone who knows how, please tell me.)

5. Use manual focus with aperture rings. (Yes, this works well. All we have to do is mount the fully manual lens to the Sigma SA mount, sometimes a tall order.)

I disagree with the sentiment (not expressly stated in the article) that these problems are only found in legacy lenses. To prove this all I have to do is find focus shift in my 24-105 f4 Art lens. I haven't seen focus shift yet, but I haven't tested this lens for it either.

Much work to do.

 

[Tom Schum]

I tried to test for focus shift. Lens is 24-105mm F4 DG OS HSM | A, 1 month old. Camera is sd Quattro running firmware v1.03.

1. Set manual focus to a given point on slanted ruler, using EVF and focus peaking.

2. Take a shot at F4 and one at F8, and others in between, without changing the focus setting.

3. Inspect images for evidence of focus shift.

I chose a focal length of about 50mm, and used a focus square to the left of center. In manual focus mode, this is where the magnifier looks. Camera was on a tripod, about 4 ft from the focus point. While I was writing this post, the cat moved into my test setup and fell asleep. So should we all...



When looking at the test images I discovered that focus is a little off even at max aperture: Focusing at a point on the slanted ruler does not give me a symmetrical focus fade out around this chosen point, but the point of focus is a little bit further away than this. Maybe it is time to get out the USB dock. On the other hand, this is not the same as the focusing I did. So, this quirk bothers me. I'm using the focus magnifier and focusing manually, yet the images show a different focus point! I repeated my attempts to manually focus properly, watching the focus peaking for symmetry around the focus point, and still got the same result. Something fishy here. I am not likely to see this using a fully manual lens, at least this is my guess at this point.

Moving on, I inspected the images and could not find anything to show a definite change in focus as F-stop was changed.

Conclusion is that even though the lens is set to manual focus with image stabilization off, something is happening that is beyond my control, and this changes the focus point.

--
Tom Schum
Every day a new image.

 

[xpatUSA]

. . . When looking at the test images I discovered that focus is a little off even at max aperture: Focusing at a point on the slanted ruler does not give me a symmetrical focus fade out around this chosen point, but the point of focus is a little bit further away than this.

Apart from the back-focusing that was observed, the non-symmetry could be because the depth of field itself is always asymmetric about the plane of focus, more behind than in front (axially speaking). Also, the ruler divisions, by definition of the ruler being angled, will have different heights in the image plane.

I hadn't thought of the foregoing when making the suggestion but it should still work for the purposes of comparison using your fixed test condition.

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[Tom Schum]

. . . When looking at the test images I discovered that focus is a little off even at max aperture: Focusing at a point on the slanted ruler does not give me a symmetrical focus fade out around this chosen point, but the point of focus is a little bit further away than this.

Apart from the back-focusing that was observed, the non-symmetry could be because the depth of field itself is always asymmetric about the plane of focus, more behind than in front (axially speaking). Also, the ruler divisions, by definition of the ruler being angled, will have different heights in the image plane.

I hadn't thought of the foregoing when making the suggestion but it should still work for the purposes of comparison using your fixed test condition.

 

[xpatUSA]

. . . When looking at the test images I discovered that focus is a little off even at max aperture: Focusing at a point on the slanted ruler does not give me a symmetrical focus fade out around this chosen point, but the point of focus is a little bit further away than this.

Apart from the back-focusing that was observed, the non-symmetry could be because the depth of field itself is always asymmetric about the plane of focus, more behind than in front (axially speaking). Also, the ruler divisions, by definition of the ruler being angled, will have different heights in the image plane.

I hadn't thought of the foregoing when making the suggestion but it should still work for the purposes of comparison using your fixed test condition.

Yes I believe you're right about the asymmetry in the depth of field.

I tried again with the 50 mm macro lens, this time the error was about the same. Then I turned off focus peaking and that allowed me to focus more symmetrically around the desired point. Lens is EX DG 50mm 1:2.8 Macro.

The focus peaking error I'm getting might have something to do with lens bokeh: different beyond the focus point vs in front of the focus point. Using focus peaking I got a backfocusing error.

Without focus peaking and with magnification I was able to get much better accuracy, maybe 4x better. With peaking the error occurs regardless of whether or not magnification is used. Maybe this has been a bigger problem for me than focus shift.

I took a couple shots with the macro lens at F2.8 and at F5.6 and again could not definitively pin anything on focus shift vs aperture. If there was any it was lost in depth of field.

You're right, this is likely impossible to nail down.

Out of interest, I tried it with the SD1M, manual focus (with viewfinder 1.32X magnifier). I used RawDigger to export the top layer so as to eliminate processing and see what the lens does:



I focused on the "6" line at the edge of the rule. But it focused apparently nearer to "4" - pretty hard to believe, that's a back-focusing of about an inch !!

Then I compared for any shift of that apparent focus at "4":



Can't see any but I'm left wondering about that back-focus before jumping into AFMA mode!

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[xpatUSA]

Yes I believe you're right about the asymmetry in the depth of field.

I tried again with the 50 mm macro lens, this time the error was about the same. Then I turned off focus peaking and that allowed me to focus more symmetrically around the desired point. Lens is EX DG 50mm 1:2.8 Macro.

The focus peaking error I'm getting might have something to do with lens bokeh: different beyond the focus point vs in front of the focus point. Using focus peaking I got a backfocusing error.

Without focus peaking and with magnification I was able to get much better accuracy, maybe 4x better. With peaking the error occurs regardless of whether or not magnification is used. Maybe this has been a bigger problem for me than focus shift.

I took a couple shots with the macro lens at F2.8 and at F5.6 and again could not definitively pin anything on focus shift vs aperture. If there was any it was lost in depth of field.

You're right, this is likely impossible to nail down.

I have a free utility called ImageJ. Among many other things, it lets you draw a line on an image and then show a graph of the pixel levels along that line. From that graph, a point of best contrast can be estimated and, if that line is crossing the marks on a ruler, the point of highest contrast should be close to the focal point:



Unfortunately, the graph slopes because of uneven lighting but I'd say the focus was at the 6" mark, which was my target. More even lighting would have helped. :-(

Hovever, in ImageJ, you can zoom in and re-draw the line:



From that graph, it appears that the focus point is closer to 5-1/2" showing us that relying only on our eyes to make these determinations is not as accurate as we might like.

The ruler is actually yellow.

The graph depends on which direct the line is drawn in. I went right-to-left in these, so the graph is backasswards to the image, sorry.

I learned from these tests that trying to accurately manual-focus the 17-70mm 'C' model when on the SD1 Merrill is virtually a waste of time compared to my live-view 10X Panasonic or even the humble DP2s for that matter. Quite irritating, that.

Minor rant: Even though the prime thing with the Foveon for me is the lack of color aliasing, what's the point if I can't focus the stupid DSLR ??

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[xpatUSA]

. . using Center Spot AF.

Not enough to be statistically significant, with only one shot at each f-number:


please view at 100% size. EXIF pop-up is for the lower shot.

Shots are X3F top-layer 16-bit TIFF exports from RawDigger. Graphing software is ImageJ.

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[xpatUSA]

Tom Schum wrote:<>

If Sigma made any manual lenses in SA mount this chore would be easy.

I used to own an older Sigma 50mm f/2.8 macro that had an aperture ring!

It was thrown in with my first SD9.

IIRC, you got electronic control from the camera by setting it to the max f-number which I believe was 32 on that model.

How about one of these (28mm):

see:

Loading…

www.imaging-resource.com

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[Tom Schum]

Tom Schum wrote:<>

If Sigma made any manual lenses in SA mount this chore would be easy.

I used to own an older Sigma 50mm f/2.8 macro that had an aperture ring!

It was thrown in with my first SD9.

IIRC, you got electronic control from the camera by setting it to the max f-number which I believe was 32 on that model.

How about one of these (28mm):

see:

http://www.imaging-resource.com/lenses/sigma/28mm-f1.8-ex-dg-aspherical-macro/review/

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

Must have been a lot less expensive to take off that aperture ring and set the aperture electrically. Now everything is under microprocessor control instead of our control.

--
Tom Schum
Every day a new image.