I have heard that Nikon Z sensors (most made by Sony?) performs great with M-lenses but have never seen actual side by side comparisons. If someone has the CV 28/2 Ultron II (or any other M wide lens) and both Nikon Z and M bodies, please post a side by side comparison showing the mid and edge areas using the CV 28/2 II at infinity distance while focusing on center. Preferentially both bodies should share a similar megapixel count.
I have posted how the 28/2 Ultron II performs on both Sony A7R IV (here) and Leica SL2 (here).
I compared both for the two M lenses I have - 35 and 28 summicron asph v1.
The 35 fair better with roughly 1 to 1.5 stop of penalty on corner performance, that lens already showing astigmatism (I presumed) on M10-R (check out photographyblog M10R review) and probably 1 stop worse performance due to pronounced field curvature.
28 OTOH having about 2 to 2.5 stops of penalty. On M it actually performed better than the 35mm but much worse on Z.
Both exhibit blue tint edges that can be easily fixed though sometimes not fully depends on lighting. Vignette wasn't an issue but it'll remain there even stopped way down.
I was pretty happy with the color rendition and rendering in general, the bokeh actually smoother on Z, but ultimately decided that I'll use the Z as my video and probably lowlight street camera when 40/2 comes out. The corner performance was visible even after scaled down (not so much on bokeh diminishing towards edges as they don't look too different than on M), otherwise it's a great body for M lenses as the ergonomics are great and IBIS worked incredibly well.
Fred Miranda wrote:
I have heard that Nikon Z sensors (most made by Sony?) performs great with M-lenses but have never seen actual side by side comparisons. If someone has the CV 28/2 Ultron II (or any other M wide lens) and both Nikon Z and M bodies, please post a side by side comparison showing the mid and edge areas using the CV 28/2 II at infinity distance while focusing on center. Preferentially both bodies should share a similar megapixel count.
I have posted how the 28/2 Ultron II performs on both Sony A7R IV (here) and Leica SL2 (here).
Speaking of the older 28mm Ultrons, the 28/1.9 LTM was one of my first MF lenses and I used it a lot with my NEX-6 (with the APS-C crop) but later when I got an A7R I couldn't quite get as nice infinity shot results with lens as I would have liked even at f8 so I sold it at some point. I did like the lens a lot for closer range shooting though. With the 28/2 VM v1 I didn't like the focus shift. I sometimes miss the 28/1.9 LTM and might add it back to my collection some day if a good copy comes by at a decent price.
Anyway, I find these small rangefinder lenses very nice and enjoyable to shoot with overall and 28mm is one of my favorite focal lengths....Show more →
The 28 Ultron LTM practically lived on my Sony. That plus a 35mm Ultron LTM. I worked around the field curvature issue for the most part with the 28. The 35 was more of a problem with the edges coming into focus when not desired. Both were superlative for near subjects (as long as the background was not too distant).
When I got the Z, with pixel peeping on the higher mp files, I found that the 28 was soft on one edge when focusing on distant subjects so I tested it on the M and Sony to see what 24mp showed, paying more attention to the edge in question, and both showed the same effect. Higher mp simply shows whatever issues exist more clearly. What got my attention with the new version is how close the lens design is to the original, and it's quite diminutive ... ; - )
Fred Miranda wrote:
I have heard that Nikon Z sensors (most made by Sony?) performs great with M-lenses but have never seen actual side by side comparisons. If someone has the CV 28/2 Ultron II (or any other M wide lens) and both Nikon Z and M bodies, please post a side by side comparison showing the mid and edge areas using the CV 28/2 II at infinity distance while focusing on center. Preferentially both bodies should share a similar megapixel count.
I have posted how the 28/2 Ultron II performs on both Sony A7R IV (here) and Leica SL2 (here).
Fred Miranda wrote:
Aside from using a M body, the best mirrorless camera to adapt M lenses is the Leica SL series because it has the thinnest packaging in front of the sensor. Furthermore these sensors are equipped with 'off-set' microlenses in front of every pixel used to direct light coming from a very steep incident angle. This avoids astigmatism and induced field curvature which is the typical result when using camera sensors with a thicker sensor stack, and that includes the Nikon Z series.
LarsHP wrote:
I feel confident that astigmatism and induced field curvature is caused by the thicker sensor glass layer. The microlenses in the M sensors seems to primarily help reducing vignetting. This is my conclusion based on comparing the M 240 to my Z6UT (and indirectly the unmodded Z6). However, as mentioned, the Z6 is a BSI sensor while the M 240 is not.
I agree. That is also my understanding. Microlenses help redirect light more straight down into deep pixel wells to the actual light sensing component. The result is reduced vignetting and edge color shift. BSI sensors have the benefit of shallower wells that reduce the need for offset microlenses.
My brief experience some years ago with the SL was that wider M lenses performed somewhat worse than on M but better than on Sony.
From the drawings I have seen of the M sensors, the taller micro lenses directs the light more into the appropriate pixel well, but my tests show that there is no color shift or smearing towards the corners on the Ultra Thin modded Z6 while the M 240 show more global color shift (Italian flag). This tells me that the micro lenses in the M sensor is not part of the equation regarding color shift when BSI sensors are used. I understand that it may be helpful with pre-BSI sensors though.
rscheffler wrote:
(...)
Microlenses help redirect light more straight down into deep pixel wells to the actual light sensing component. The result is reduced vignetting and edge color shift. BSI sensors have the benefit of shallower wells that reduce the need for offset microlenses.
My brief experience some years ago with the SL was that wider M lenses performed somewhat worse than on M but better than on Sony.
The M240 was designed with shallower wells, but I don't believe it's an actual BSI sensor. For example see here where it's stated the conducting elements above the photodiodes were made of thinner copper to bring the microlenses closer to the photodiodes. In a BSI sensor the conducting elements are behind the photodiodes. I know I still get color shift with specific, more problematic lenses on the M240, but it's better than it was with the M9. The Z6's sensor is BSI from what I've read, which would give it an advantage regarding color shift.
Another factor for color shift is likely the IR blocking filter over the sensor. With wider short exit pupil distance lens designs with light reaching the sensor at more oblique angles, those off-axis light rays travel farther through the IR filter, resulting in a stronger filtering effect, which in turn results in uneven color across the image.
rscheffler wrote:
The M240 was designed with shallower wells, but I don't believe it's an actual BSI sensor. For example see here where it's stated the conducting elements above the photodiodes were made of thinner copper to bring the microlenses closer to the photodiodes. In a BSI sensor the conducting elements are behind the photodiodes. I know I still get color shift with specific, more problematic lenses on the M240, but it's better than it was with the M9. The Z6's sensor is BSI from what I've read, which would give it an advantage regarding color shift.
Another factor for color shift is likely the IR blocking filter over the sensor. With wider short exit pupil distance lens designs with light reaching the sensor at more oblique angles, those off-axis light rays travel farther through the IR filter, resulting in a stronger filtering effect, which in turn results in uneven color across the image....Show more →
Hi Ron,
I know that with the Sony A7rII and A7rIII cameras, these cameras had much less problem with color issues with many of the M mount wide angle lenses like my Zeiss 35mm f1.4 Distagon ZM than my original A7r cameras. Even after Kolari Vision performed a thin sensor modification on my one A7r V3 modified camera I would need to use corner fix with that lens and my Leica M 24mm f3.8 Asph lens. I stopped using those lenses with that camera and these lenses no longer cause a color issue with my Kolari Vision UT sensor modified A7rII (need new color calibration however). Many of us who had not had a thin/UT sensor modification with an A7rII or A7rIII were able to get by with the Zeiss 35mm f1.4 ZM for color though the lens would need much more stopping down on the stock camera bodies to perform acceptably.
I am not Ron, but since I tested my 28mm Summicron-M Asph II on a Leica M 240, as mentioned a few times, I can confirm that there was quite visible color shift (often called the "Italian flag" issue, which shows as cyan in the left and red i the right hand side of the image, when white balance is set in the center).
However, this is of course without lens coding in the Leica M camera. If you select a (Leica) lens profile which fits the characteristics of your Voigtländer lens, then the color shift (as well as some vignetting and possible distortion) will be corrected when you open the file in your editing software. With the ultra-thin modified Nikon Z6, the color shift is less, but there is no in-camera profile available, so the correction needs to be done by the user in the editing software. This is no big deal since you can make and save such a profile and select it when needed.
al404 wrote:
In case I decide to get an M9 or M240 can I have issue on these camera with latest Voigtlander lens?
Regarding BSI sensors compared to the previous design, I found the below, where the difference in pixel well opening appears quite big which suggests the superiority of the BSI technology regarding steep angles of light.
LCT @ DPReview posted links to the two images below, both shot with the Leica 21mm f/3.4 Super-Angulon lens @ f/5.6 and on Leica M 240 versus a Kolari Vision Ultra Thin sensor glass modded Sony a7RII.
I did find notice about this kind of issue on some PRE digital lens
Voigtlander 15mm v1 and 2 has this kind of issue, but seems that on A7iii is less pronounced respect early cameras
Voigtlander 15mm v3 does not issue, as far as I know is the same for Voigtlander 21mm f4 ( not good for digital ) and 21mm f3.5 seems to be fine on mirrorless
that is why I was pretty surprised of the issue on this new 28mm
The reason why the Voigtländer 21mm f/3.5 and 15mm f/4.5 III work well on mirrorless cameras is because they are retro focus design and made in both M and E mount. The 28mm Ultron is only made in M mount and isn't a retro focus design (AFAIK).
al404 wrote:
I did find notice about this kind of issue on some PRE digital lens
Voigtlander 15mm v1 and 2 has this kind of issue, but seems that on A7iii is less pronounced respect early cameras
Voigtlander 15mm v3 does not issue, as far as I know is the same for Voigtlander 21mm f4 ( not good for digital ) and 21mm f3.5 seems to be fine on mirrorless
that is why I was pretty surprised of the issue on this new 28mm
al404 wrote:
In case I decide to get an M9 or M240 can I have issue on these camera with latest Voigtlander lens?
I don't have enough first-hand experience with the latest Voigtlander lenses to answer this conclusively, but from the photos I've seen posted online, there shouldn't be a problem with the 'Italian flag' type of color shift.
From my experience with the M9 and M240, if a lens were to cause such a color shift, it would be worse with the M9.
---------------------------------------------
al404 wrote:
I did find notice about this kind of issue on some PRE digital lens
Voigtlander 15mm v1 and 2 has this kind of issue, but seems that on A7iii is less pronounced respect early cameras
Voigtlander 15mm v3 does not issue, as far as I know is the same for Voigtlander 21mm f4 ( not good for digital ) and 21mm f3.5 seems to be fine on mirrorless
that is why I was pretty surprised of the issue on this new 28mm
The only 'pre-digital' lens I currently have at hand that I know causes 'Italian flag' color shift problems is the Voigtlander 12/5.6 LTM and the problem is less severe and easily fixable in Lightroom with M240 files. With the M9 I had to make Corner Fix profiles for this lens as well as the VM15/4.5 II, ZM21/2.8 and ZM35/2.8 (none of which I still own). Indeed the VM15/4.5 III is much better in respect to minimizing color shift. The 21/2.8 and 35/2.8 (on the M9) were less severe than the ultra-wides and in many situations the shift seen with the 35/2.8 was hidden by image content. This was if I set a correction profile in-camera to already reduce the color shift problem.
I owned the Leica 28/2 v1 for many years images from it, with auto lens coding enabled in-camera, still showed some signs of edge color shift in the corners. This was both on the M9 and M240. It was most noticeable in very monochromatic scenes, such as snow, concrete, etc. I've since moved to the 28 Lux and it is much better in respect to minimal color shift.
So why is the new VM28/2 problematic? I suspect it's party the reason Lars suggests. Its compact size might mean the exit pupil distance isn't long enough. It exhibits fairly strong vignetting, which indicates relatively strong off-axis oblique ray angles. As a result, I think there are a couple sensor related factors at play that we might be mixing up. With some sensors, particularly non-BSI, it might result in some Italian flag type color shift. But given it's a recent lens design, I would expect this to be minimal with more recent cameras. The other, as suggested in this post by Sean Read about using Adobe's Flat Field correction software, is it's likely caused by the sensor's IR blocking filter.
The nature of color shift I've noticed in sample images from the new VM28/2 tends to be cyan/blue, which is probably more due to the oblique ray angle through a given sensor's IR filter than the underlying sensor's physical design/structure. I also see this with my VM35/1.7 if in-camera lens coding is not applied - images are noticeably cooler/bluer than with coding.
In Sean Reads post above, he mentions seeing such color shift also with a couple Sony cameras and specific Sony-mount lenses, despite one of the cameras being a BSI sensor design. This to me indicates color shift can be a combination of at least two sensor related factors: sensor type (BSI vs. non-BSI) and off-axis light ray interaction with the sensor's IR filter.
LarsHP wrote:
If these are not fake, I think this settles the discussion regarding how much the micro lenses contribute to color shift...
Those are great examples of the Italian flag problem. IIRC the 21/3.4 Super Angulon is one of the worst offenders. But I'm not convinced microlenses don't improve the result. Maybe it's not as significant a factor relative to physical sensor structure/design, such as BSI vs. non-BSI.
Aug 08, 2021 at 11:35 AM
Steve Spencer Offline • • • • • • • Upload & Sell: On
The only 'pre-digital' lens I currently have at hand that I know causes 'Italian flag' color shift problems is the Voigtlander 12/5.6 LTM and the problem is less severe and easily fixable in Lightroom with M240 files. With the M9 I had to make Corner Fix profiles for this lens as well as the VM15/4.5 II, ZM21/2.8 and ZM35/2.8 (none of which I still own). Indeed the VM15/4.5 III is much better in respect to minimizing color shift. The 21/2.8 and 35/2.8 (on the M9) were less severe than the ultra-wides and in many situations the shift seen with the 35/2.8 was hidden by image content. This was if I set a correction profile in-camera to already reduce the color shift problem.
I owned the Leica 28/2 v1 for many years images from it, with auto lens coding enabled in-camera, still showed some signs of edge color shift in the corners. This was both on the M9 and M240. It was most noticeable in very monochromatic scenes, such as snow, concrete, etc. I've since moved to the 28 Lux and it is much better in respect to minimal color shift.
So why is the new VM28/2 problematic? I suspect it's party the reason Lars suggests. Its compact size might mean the exit pupil distance isn't long enough. It exhibits fairly strong vignetting, which indicates relatively strong off-axis oblique ray angles. As a result, I think there are a couple sensor related factors at play that we might be mixing up. With some sensors, particularly non-BSI, it might result in some Italian flag type color shift. But given it's a recent lens design, I would expect this to be minimal with more recent cameras. The other, as suggested in this post by Sean Read about using Adobe's Flat Field correction software, is it's likely caused by the sensor's IR blocking filter.
The nature of color shift I've noticed in sample images from the new VM28/2 tends to be cyan/blue, which is probably more due to the oblique ray angle through a given sensor's IR filter than the underlying sensor's physical design/structure. I also see this with my VM35/1.7 if in-camera lens coding is not applied - images are noticeably cooler/bluer than with coding.
In Sean Reads post above, he mentions seeing such color shift also with a couple Sony cameras and specific Sony-mount lenses, despite one of the cameras being a BSI sensor design. This to me indicates color shift can be a combination of at least two sensor related factors: sensor type (BSI vs. non-BSI) and off-axis light ray interaction with the sensor's IR filter....Show more →
Hi Ron,
Thanks for this post. I hadn't heard about the interaction with the sensor's IR filter, but that makes sense.
That's one of my points in a recent post: Micro lens design probably have an impact in itself, but BSI sensor design versus pre-BSI design appears to be more important regarding color shift / the "Italian flag" syndrome.
rscheffler wrote:
Those are great examples of the Italian flag problem. IIRC the 21/3.4 Super Angulon is one of the worst offenders. But I'm not convinced microlenses don't improve the result. Maybe it's not as significant a factor relative to physical sensor structure/design, such as BSI vs. non-BSI.
Today I finally received my 28mm Summicron Asph II from Wetzlar.
I did a quick vignetting comparison and unsurprisingly the Ultron II vignettes considerably more wide open than the 'cron. The Ultron II still vignettes more @ f/2.8, but @ f/4 and f/5.6 the Ultron II vignettes less!!!
Another thing this test shows is that the Ultron II seems more even across the frame in terms of color tint. In other words, the Ultron II has less "Italian flag" issue than the Summicron. ( EDIT: I WAS WRONG ABOUT THE ITALIAN FLAG ISSUE. See this post: https://www.fredmiranda.com/forum/topic/1704868/18#15674849 )
All the above is on my ultra-thin sensor glass modded Nikon Z6. Results on Leica M cameras might be different.
LarsHP wrote:
Today I finally received my 28mm Summicron Asph II from Wetzlar.
I did a quick vignetting comparison and unsurprisingly the Ultron II vignettes considerably more wide open than the 'cron. The Ultron II still vignettes more @ f/2.8, but @ f/4 and f/5.6 the Ultron II vignettes less!!!
Another thing this test shows is that the Ultron II seems more even across the frame in terms of color tint. In other words, the Ultron II has less "Italian flag" issue than the Summicron.
All the above is on my ultra-thin sensor glass modded Nikon Z6. Results on Leica M cameras might be different. ...Show more →
Hi Lars,
Thank you for the additional information. It is beginning to look like the sweet spot for the Voigtlander 28mm f2 Ultron II Asph VM lens from your testing, Fred's testing, my testing, and others is around f4 to possibly f5.6. Wider apertures have their uses as does stopping down more for greater DOF for landscape and other work. The lens is a great buy for what it offers and is a major competitor to the Leica 28mm f2 Asph lens which is more than 5X the cost.
One thing we could say about the 28mm Ultron Asph II regarding its vignetting is that, apart from the very center, it doesn't fully behave as an f/2 lens, but more like an f/2.4 lens. The vignetting wide open on the Z6UT is quite dramatic, while the 'cron has a more moderate vignetting.
I don't know the vignetting difference in f-stop between the two, but in practice I would use the 'cron in some situations without vignetting compensation at all, while I wouldn't do so with the Ultron II (unless for a creative effect).
