Corner Color Drift on M10?

[cookedart]

Advertisement (gone after registration)

Has anyone read any reports on how the m10 performs with respect to corner color shift on wide angle lenses?

[nico1974]

Has anyone read any reports on how the m10 performs with respect to corner color shift on wide angle lenses?

 

I had the opportunity to test it out on Friday with a Voigtlander 21/4 lens and I could not see any corner colour shift with no profile selected. I did not have an SD card to download the image and inspect further on my calibrated monitor but did not notice anything on the LCD screen. Would be very curious for someone to test this out properly until my Leica M10 arrives

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[cookedart]

Thanks, even first impressions are good to hear!

[steppenw0lf]

Sean Reid (reid reviews) did some comparisons to other cameras/sensors. (regarding high ISO etc.)

If he had found any special corner weakness he would have written about it.

[cookedart]

I've read all of Sean Reid's stuff on the m10. He doesn't yet get into this issue, at least to the depth I'm looking for, hence me asking for impressions here. It's not so much about weakness as what improvements have been made vs the m240. Lots of older, compact wides have a notable amount of color shift.

[steppenw0lf]

I expect no/only small improvements regarding this color shift. So I use/used these lenses only on the M246 where this is no problem - or on any other digital M used for b&w.

But since I have the WATE this is no issue anymore. (have not used the older lenses anymore)

It looks as if the M240, M246 and M10 might have a sensor from the same source (no official statement from Leica, but users found some common peculiarities). Accordingly I assume that the "problematic" behaviour of these older lenses will be similar on all three.

If there had been big improvements, then probably marketing would have noticed us about them.  

[UliWer]

Advertisement (gone after registration)

Well, that's an assumption, since marketing would rather be shy to talk about issues of models they still sell.

 

Since the M10-Sensor is approx. 3 years "younger" than the sensor for the M240 I expect significant progress in this field. If the new camera was only on the same level in coping with the issue, that would be a serious dealbraker for me. The M9 - with the last firmware - was already better than the M240, though it wasn't perfect.

[steppenw0lf]

Leica is making steady progress, but usually lenses and sensor have to fit together well. Leica lately changed the M 28mm lenses to be better prepared for the new sensors. (refresh of 2.8 and 2.0 28mm and addition of the exceptional 1.4/28mm, also refresh of the 2.0/35mm)

So there is enough progress to applaud.

But the older (actually very old) WA lenses with color shift (all from the film era) are still unchanged, so progress is certainly limited. (and not necessarily a reason for disappointment    )

 

And even today you can (with software correction) try to get rid of the false colors. But usually the results are "mixed".

[UliWer]

There are some hints from Leica:

 

In the interview with the people who were responsible for the M10 on the Leica Webside you find the following statements:

 

"Henning Rafflenbeul: The M10’s sensor also features micro-lenses that have been specifically optimised for the optical characteristics of M lenses. This continues to ensure the best image quality within the M-lenses’ compact design.

Jesko von Oeynhausen: Also, side effects such as vignetting or colour shifts on the edges of the image have been reduced even further with the sensor of the M10."

 

http://blog.leica-camera.com/2017/01/18/looking-back-move-forward/

 

Let's wait and see, if there is real progress.

Edited January 24, 2017 by UliWer

[adan]

While I always believe in "Trust but verify," and look forward to seeing what my 21 non-ASPH (my money lens) does on the M10, I am not overly worried. Except for Sandymc's blog post about the M10 firmware "seeming rushed" in some ways.

 

The original M9 firmware had serious Italian-flag/red-edge problems with this troublesome superwide lens, but after a year or so Leica got the corrections fixed with a firmware upgrade.

 

The M240 handled that lens nicely right out of the box, however, so I think Leica is paying close attention to this detail these days.

 

At least for supported 6-bit lenses. I doubt the old Super-Angulons or C/V 15mm v. 1/2 will ever work very well for color.

[uhoh7]

I had the opportunity to test it out on Friday with a Voigtlander 21/4 lens and I could not see any corner colour shift with no profile selected. I did not have an SD card to download the image and inspect further on my calibrated monitor but did not notice anything on the LCD screen. Would be very curious for someone to test this out properly until my Leica M10 arrives

With latest firmware on M9 coded as 21 Elmarit pre-asph, the CV 21/4 does not shift color that I can see. Softness on the edge, yes.

 

Panoplay by unoh7, on Flickr

 

L1054387 by unoh7, on Flickr

 

L1054327 by unoh7, on Flickr

 

I'm also very interested if M10 will do better with this lens and other wides, beyond colorshift, better in the edges and corners. EG ZM 21/4.5, Leica 21 v1 and 21/2.8 pre-asph. Also 28 skopar and CV 15 v1 and 2

 

I do have SEM21, but this skopar is so easy to grab when I'm on a bike or going very light. I don't worry about dropping it, either. It's made me some pretty decent shots

Edited January 24, 2017 by uhoh7

[cookedart]

Well, that's an assumption, since marketing would rather be shy to talk about issues of models they still sell.

 

Exactly, I don't want to stay in the realm of assumptions, want actual verified tests. One thing that probably encourages me is that Leica seems to be focusing the camera on M glass, as opposed to the M240 which was supposed to be a shooting platform for both R and M lenses (which is now relegated to the SL),  and its microlens arrangement was designed as such. Hoping to hear good progress in this regard.

[Rick]

uhoh7 - Nice shots.  Big Wood in the fall?

 

Rick

[steppenw0lf]

Exactly, I don't want to stay in the realm of assumptions, want actual verified tests. One thing that probably encourages me is that Leica seems to be focusing the camera on M glass, as opposed to the M240 which was supposed to be a shooting platform for both R and M lenses (which is now relegated to the SL),  and its microlens arrangement was designed as such. Hoping to hear good progress in this regard.

 

The R lenses were never in need of special microlens placement - this was only/mainly done for the M lenses.

It's obvious since the R lenses work almost anywhere, even with Sony cameras (or CaNikon), that are "difficult" with some compact M lenses.

Edited January 25, 2017 by steppenw0lf

[cookedart]

The R lenses were never in need of special microlens placement - this was only/mainly done for the M lenses.

It's obvious since the R lenses work almost anywhere, even with Sony cameras (or CaNikon), that are "difficult" with some compact M lenses.

 

Sorry but this isn't exactly true. While you're right that it doesn't need special placement, Leica still designed the microlens array in such a way that it wouldn't compromise R lenses.

 

http://www.the.me/the-leica-m-max-sensor-explained/

 

Consider the quote

 

It wasn’t easy to find a partner who could fulfill our specific requirements for the sensitivity of the sensor in relation to the incident angle because of the M’s low flange distance, considering the quality of the color filter and much more. With the M9 we were still working with microlens shifting to get a grasp on the problem, but with the M it was clear right from the beginning that the camera had to also work with the lenses of Leica’s R system, which are much more telecentric (…) The second criteria was the energy consumption, because with live view and video the sensor needs to run continuously.  

 

Essentially, Leica had been using offset microlenses in the m8 and m9, and went to a different solution altogether with the m typ 240 since the offset microlenses would hamper R performance. Their solution was to try to reduce the distance between the photo diode and the colour filter, as well as reshape the microlenses to handle the incident angles of the light rays. 

 

 

My point is, on the m10, they are likely more free to try more options that would be potentially harmful to R lenses, in favor of M lenses.

[uhoh7]

@Rick TY for kind words, yes Bigwood River about a mile from my house

 

Sorry but this isn't exactly true. While you're right that it doesn't need special placement, Leica still designed the microlens array in such a way that it wouldn't compromise R lenses.

 

http://www.the.me/the-leica-m-max-sensor-explained/

 

Consider the quote

 

 

Essentially, Leica had been using offset microlenses in the m8 and m9, and went to a different solution altogether with the m typ 240 since the offset microlenses would hamper R performance. Their solution was to try to reduce the distance between the photo diode and the colour filter, as well as reshape the microlenses to handle the incident angles of the light rays. 

 

 

My point is, on the m10, they are likely more free to try more options that would be potentially harmful to R lenses, in favor of M lenses.

The whole subject is very interesting and I'd love those familiar with sensor options, and digital M(x) performance with really high ray angles to share their knowledge.

 

The CV21/4 is a great benchmark, as so many have the lens, so testing on various platforms is easier. It has as severe issues as almost any, though cv15 v1 and 2 are worse yet. ZM 21/4.5 is the lens which would be nice to have really working well, but doesn't yet seem to, but check me there.

 

The modern UWAs are much easier. For example, Jim Kasson did detailed tests with A7ii.Kolari vs M240 with a number of M wides. First, let's be clear on the known differences, which have to do with coverglass and filterstacks. Leica does not use filter stacks at all, unlike Canikon or Sony, which use a clear hard .7mm coverglass glued to the main sensor surface, glue is on edges. Then over this they put IR cut, AA (or not), and other filter types: that's the "stack". It varies in thickness from about 1.1mm to 1.9mm, the former being Canikon, and the latter Sony. There is some further variation, in particular models, but this gives the general idea. Leica has none of that at all. They do all this in the coverglass itself, which is "normally" clear. For M8 it was .5mm of schott S8612, a very high performance IR cut, which is also a bit sensitive to corrosion. That was a bit lean for IR, hence the IR cut filters people added to lenses. But it is the thinest glass ever over a large sensor, and explains much of the striking clarity of the M8. Those images really dropped my jaw, many in BW, and in retrospect, despite the controversy over IR and the crop, the incredible M8 performance established Leica in the digital world indelibly. Then came M9, which used S8612, but now at .8mm. Results were spectacular to the point that the camera to this day is fantastic at base ISO. It is still in very wide use today, as everybody knows, and some not stupid people prefer it. Perhaps the choice of coverglass was a bit too high-performance, as the sensitivity to humidity, which was always known, has become an issue for some people, as we all know. I'm still on that sensor, and I hope it never goes, because it's better than the new one in terms of IR cut performance. The replacement is apparently BG55, also at .8mm. It is very tough in harsh and humid environments, but has roughly 10% lower transmission. Whether that is a real world difference, I'm not sure anyone has looked into closely. The coverglass on M240 is unclear but we can assume it's also about .8mm thick. Why is that thickness such a big deal?

 

Enter Sony A7, which many of us hoped would be great with RF glass. As we all know it's not and a big factor is there is literally an extra 1.8mm of extra glass over the sensor to bend the steeper rays of RF wides. "Smearing", is actually an FC in most cases, and you can focus a Sony on the edge and get them quite good, but the center goes. This suggests right away there is not a microlens problem with the Sonys. If there was the edges would be softer, when POF was on them. The ZM 35/1.4 has great edges on stock sony A7rii wide open! But no centers It's one or the other. BTW that lens is the best 35 which can go on the r2, and beats everything for landscape by F/5.6ish, native or not. Those guys are putting special filters to compensate for the FC in the faster stops, and modding the adapters and the lens to do it. Fred Miranda, himself, is all over this:

http://www.fredmiranda.com/forum/topic/1453834/25

 

OK, more to our point, is the question, what would happen if you just did the Leica thing and glued .8mm over the A7rii sensor and dumped the filter stack? Here is the answer:

https://flic.kr/p/PAGgSy

You might to check out Sam's recent photo stream there. He's at the cutting edge of what you can do with a Sony. He has shots with various lenses. WB issues are there, but I have used the RAWs and they are easily corrected in LR. The current Kolari mod to A7x (not what Sam is using), keeps the .7mm coverglass and reduces the stack from 1.9mm to about .8mm and does equal the M240 with some lenses, like the SEM 18 and 28 Lux. Lloyd also verified this. But Kasson shows, and I concur (I have A7.kolari), not quite with many others, like SEM24, ZM35/2 etc. Way better than stock, though. In general images have improved clarity, especially with the A7 and A7ii which loss the AA filter in the Kolari mod. But Sam, with A7rii, has his tech go much further, and actually remove the coverglass in a clean room, remove the filter stack and replace with a single .5mm IR cut. He manages to seal against dust. This is one guy in Taiwan. Sam recommends to use a new choice of IR cut closer to Sony AWB, and .7mm. This camera is likely to be very close to M240 performance with steep rays. There is a Sony app to make profiles for the problem lenses, like we are discussing. Let's consider Leica Profiles, to show what they can do and might come into the mix of helping wides.

 

Here is CV 21 on M9 without profile, today:

L1056689 by unoh7, f/4

 

Now with the in camera 21 Elmarit 2.8 pre-asph profile:

 

L1056688 by unoh7, on Flickr

 

You can find the fulls on flickr. Shot in Raw uncompressed, imported to LR, no editing or LR lens profiles, then exported as jpeg to flickr. I do not know the effect of such huge correction on edge noise, as I have not looked really close, but feel free to comment. The lens vignettes almost as much at F/8 and is equally corrected. This is last firmware for the "old" M9 sensor, which only differs in coverglass I believe. They did a pretty good job on the profile: but I'd almost prefer they'd left the vignette and just dealt with colorcast, so I can take it out to my taste later without the noise hit which has to come with such a push.

 

I'd love to see CV 21/4 on M240 and M10: with profile at f/4 and without. In my stream you can see F/8 also.

 

Now, as to M not having to cater to tele-centric R and better than SL, but SL better with R: Is this even true? Is the SL at base ISO better than 240 with R lenses? Any of them? I'd like to see any samples which prove this. R was designed for film, so unlike current Sony natives, there should be no need for "glass in the path" as must be used in optical bench tests for the Sony lenses to hit their marks. Ray angles on M lenses vary hugely. The 135 APO I'd very much doubt has steeper ray angles than many SLR lenses. In fact lloyd has public charts which show the ZM50/2 with a very mild ray angle not far off the Milvus 50/2, so I'd assume there are R wides which are in the same range. This would imply the M sensor must handle mellow angles also with some M glass. Please correct me here. This is why I'm skeptical the SL difference is to help the R glass. There may be some other reason....correcting natives? Additional filters for video moire, or the like, which mean the sensor is built different? SL vs M240, maybe somebody has links to careful comparison with various lenses?

 

Of course, the proof is in the pudding, and that's the real evidence we can go by. M10 will be so interesting to see with all the usual suspects, including the 28 cron and 50 Lux, which can seem like different lenses on M9 and 240. Editing is a big factor, but 50 Lux love never recovered: sure it is sharp on the 240, as is the 28 cron, but character seems quite different in so many faster shots. Do I have that wrong? How will they shoot on M10?

Edited January 26, 2017 by uhoh7

[cookedart]

Yes I can concede I'm not really sure what improvements have been made to the m10 sensor in this regard, if any. Or if R performance is better or worse. They haven't hilighted the new sensor in the same way they did the MAX sensor in the m typ 240. So it will be just guessing until we get more tests back.

 

My assertion of the SL was that it's being positioned more as a multipurpose camera, one that can take M, R, and native SL lenses and have reasonable performance with all. But I have definitely seen evidence that some m lenses don't perform as well on the SL as they do on the m 240. This suggests to me the sensor design is either a compromise or a design that favors their SL glass. I can only assume this design would favor R glass as well, but I haven't seen much assertions either way. But with leicas new focus with the m10 it's possible they could design the sensor layout to be more m centric, like the m9, which had a sensor layout that was totally unconcerned with adapting non native lenses. But that's what I hope tests will confirm or deny.

[steppenw0lf]

The conditions are the same for M lenses with a larger distance of the last element to the sensor, as for any other lens with this characteristic. Typically most SLR lenses have this characteristic.

So either all of these lenses are compromised, or none.

And so if Leica is designing the sensor for M lenses, it will automatically make it fit for many other lenses.

What you added is just the normal documentation, nothing new. And no indication of a "special treatment" for R lenses. 

[adan]

Whoa, uhoh! Stop and catch your breath!

 

But your point about microlenses and long-back-focus SLR lenses (not much impact) makes some sense.

 

If the light is already coming in telecentrically - from directly in front, mostly, instead of at a slant angle - then (referring to cookedart's diagram, since we have it), light from SLR lenses does not hit the microlenses at much of an angle on the side, but directly on the top of the "dome", and thus refracts less, so the microlenses are effectively flat glass for the purposes of light from such lenses.

 

With confirmation being that I've used just about every Leica M lens longer than 50mm on the M9, without any corner smearing or other fuzziness or color drift, and with or without lens recognition turned on. The "rangefinder wide-angle" microlenses on the M9 sensor handle those "SLR-like" lenses with no noticeable effect.

 

Keep in mind that many of Leica's telephotos are exactly the same lens whether on M or R. The various 135 f/2.8s, and the 90 APO (which was ported to an R mount just before the R line was discontinued), and the last 90 Elmarit-M, which was the 1980 R lens design, mounted for the M in 1990.

 

Now, possibly, a really long R lens might be affected by rangefinder-friendly offset microlenses. WLaidlaw can probably tell us - he used the long Visoflex (original) lenses on his M9, I believe. Up to 400mm, I think. He'd know if he saw any issues, most likely corner darkening rather than color drift.

Edited January 26, 2017 by adan

[Advertisement]

Hallo,

Look here: Leica M

[steppenw0lf]

There are also lenses - e.g. the  Summilux M 28 - that work better on the SL than on the Ms. (Maybe different now with the M10? )

It would be interesting to see  if these lenses are now improved on the M10 sensor.

And the difference between the older 2.8 and 2.0 28mm and the newer would also be interesting, maybe also for the 2.0 35 (older vs new).

These lenses were "refreshed" for the future sensors, so probably also for the M10 sensor.

Edited January 26, 2017 by steppenw0lf