M9 and Red Edge Redux--Part II, with 28mm Elmarit?

[mjh]

Advertisement (gone after registration)

You forget to take into account that the microlens layer is not necessarily perfectly positioned on top of the sensor layer. If, in assembly, the microlenses are offset even a single micron to the left or to the right, compared to the ideal position, it creates an asymmetry that would be significant to the result. Almost certainly, this is the asymmetry that the Leica representatives have in mind, even though they are reluctant to pinpoint it explicitly, as admission that it is a sample-variation issue might cause buyers to demand sensor replacement, in order to get an "ideal" sample.

Actually I had thought of this. I don’t think that the microlens pattern might not fit the pixel pattern underneath – this should be well within the precision achievable. Microlens-shifting might be different matter, though. I’m not really knowledgeable about the way the shifted pattern of microlenses is created, but maybe the amount of shifting isn’t exactly symmetrical?

 

On the other hand, the Italian Flag phenomenon was first observed with MF backs that generally don’t even have microlenses (there are some exceptions).

[Advertisement]

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

[UliWer]

I don't know if this might give a clue to anybody (it doesn't do for me) or if it just makes things more complicated:

 

Since the last firmware, when Leica tried - unsuccessfully - to do something about the red edges, one could see several examples where the red edges seemed to have been shifted from the left to the upper right corner.

[jaapv]

It was not that unsuccessful - for many lenses -not all admittedly- the problem was eliminated, or at least mitigated.

[jaapv]

Actually I had thought of this. I don’t think that the microlens pattern might not fit the pixel pattern underneath – this should be well within the precision achievable. Microlens-shifting might be different matter, though. I’m not really knowledgeable about the way the shifted pattern of microlenses is created, but maybe the amount of shifting isn’t exactly symmetrical?

 

On the other hand, the Italian Flag phenomenon was first observed with MF backs that generally don’t even have microlenses (there are some exceptions).

And a bad offender was the Kodak DCS 14 . Not a Kodak sensor. It seems these are all cameras without AA filter.

[OttoG]

Actually I had thought of this. I don’t think that the microlens pattern might not fit the pixel pattern underneath – this should be well within the precision achievable. Microlens-shifting might be different matter, though. I’m not really knowledgeable about the way the shifted pattern of microlenses is created, but maybe the amount of shifting isn’t exactly symmetrical?

 

On the other hand, the Italian Flag phenomenon was first observed with MF backs that generally don’t even have microlenses (there are some exceptions).

 

Surely it is not impossible to mount the microlens layer with, for practical purposes, perfect matching to the photosite layer in every sample. But it might be prohibitively expensive, or there may be other problems that require a compromise to be made. For example, with regard to the "sensor-cracking" issue, there has been a lot of discussion of stress issues when bonding sensor layers with each other. There may be a tradeoff between mounting methods that are precise and ones that are resilient to mechanical stress.

 

In other words, it could be that it would be easy to consistently mount the microlenses in the ideal position if the sensor were to be used under controlled temperature conditions and without exposure to vibrations, but in a camera for outdoors use, it might be difficult and expensive.

 

With regard to medium-format backs, refraction issues may affect microlens-free sensors, too, if the Bayer filter is offset from the ideal position. (The amount of light that is cancelled due to hitting two adjacent color filters, and possibly due to other color-matrix-related effects, differs depending on the angle after refraction.) But the issue is of course much less pronounced, which is mirrored in claims that microlens-free MF sensors are much better suited to non-retrofocus wide-angle lenses and lens movements than the ones that have microlenses.

[adan]

Does Kodak use the common "thermal-flow" method of producing microlenses for the M8/M9 sensors? Seems like some here are assuming the microlenses are an embossed sheet of material "attached" to the sensor somehow, whereas in most cases microlenses are produced "in place" on the sensor by the same lithographic etching techniques used to sculpt the silicon.

 

Coat the sensor with an optical material via gas diffusion or some similar process, photo-engrave it into separate squares above each pixel, then heat the material so that it melts and slumps into domes under pressure and surface tension. As depicted in these patent drawings:

 

Method for manufacturing CMOS image sensor having microlens therein with high photosensitivity

[thighslapper]

Advertisement (gone after registration)

Does Kodak use the common "thermal-flow" method of producing microlenses for the M8/M9 sensors? Seems like some here are assuming the microlenses are an embossed sheet of material "attached" to the sensor somehow, whereas in most cases microlenses are produced "in place" on the sensor by the same lithographic etching techniques used to sculpt the silicon.

 

Coat the sensor with an optical material via gas diffusion or some similar process, photo-engrave it into separate squares above each pixel, then heat the material so that it melts and slumps into domes under pressure and surface tension. As depicted in these patent drawings:

 

Method for manufacturing CMOS image sensor having microlens therein with high photosensitivity

 

I assume it is the photolithographic method..... all the lenses have to be slightly different and presumably this is the only way you can do it...... I was also a bit surprised at how expensive the process is ..... there is s lot of info here.... SUSS MicroOptics - Our MicroOptics Set the Standards. .... and what extensive uses this technology has...

 

Variability of registration of the lenses over the bayer grid may explain the camera to camera variability.

 

As a technique of obviating the peripheral light loss from very wide angle lenses it was only ever going to be very approximate.... I suppose we ought to be grateful that Leica have managed to get it to work as well as they did.

[mjh]

And a bad offender was the Kodak DCS 14 . Not a Kodak sensor.

Also not a CCD btw.

 

It seems these are all cameras without AA filter.

Indeed they are. An antialiasing filter would tend to mitigate angle-of-incidence issues.

[jaapv]

Yes, but this observation puts the red-edge problem in the same class as the moire problem the M9 has without doubt - and which is generally accepted as an unavoidable tradeoff for the ultimate image quality.

[tgray]

On the other hand, the Italian Flag phenomenon was first observed with MF backs that generally don’t even have microlenses (there are some exceptions).

 

I think it's also evident on some broadcast (and probably non-broadcast) video cameras with larger chips. I believe it's called white shading. And the cameras often have a function to help map it out. At least that's what I was told about 5 years ago.

[albireo_double]

Your friend hoppyman here

My Summicron 28 ASPH does not exhibit any red edge at any aperture in any shot I have made either. Neither my Elmar 24 nor that one did even with the first M9 firmware either (as shown in my earlier sample). I tried an Elmarit 24 and the Summicron 35 ASPH and the new Summilux 35 ASPH as well. A friend brought the CV Ultron 28 around to try too. That one also behaves as do my wides on his M9, although we haven't shot extensively in those swap arounds.

I think you must have bad karma from a previous life or something

 

Just one final diagnostic thought. You never answered which Raw processer you are using? Also any chance you are applying some CA correction setting in your processing?

 

Then you are very lucky since my 28mm Summicron shows the "Italian flag" effect in spades. The guy who sold me the lens says he never experienced the problem on his M9. It drives me crazy. I have read suggestions that the front element may be out of alignment on the lens but I doubt it. Makes the whole proposition of the M9 being a "wide-angle specialist camera" rather dubious, in my mind. I use LR 3.3. as my raw processor.

[albireo_double]

Yes, but this observation puts the red-edge problem in the same class as the moire problem the M9 has without doubt - and which is generally accepted as an unavoidable tradeoff for the ultimate image quality.

 

I agree with this observation; I am rather disappointed with the frequency of moire in my M9 images - it shows in ANY finer structures, such as tree branches, reeds, aerials, etc. It may not be evident in prints but it for sure is evident in 100% on screen magnification. What surprises me is that the problem is much more severe with the M9 than with the M8 which I used to have (and which I believe uses the same sensor, just smaller), or with my 33MP Sinar digital back (moire shows there also, for sure, but the dedicated raw processor Captureshop is typically quite good at eliminating it specifically in those cases I mentioned earlier, i.e. tree branches, aerials etc.). The combination of moire, severe "red edge" (with my 28 'Cron asph) and constantly reappearing drops of shutter lubricant on the sensor of my new M9 really make me wonder if this is an adequate price to pay for the "ultimate image quality" (and I am not even mentioning the virtual impossibility to buy ANY lenses, in any shop, these days, due to Leica not being able to meet demand...). I would have expected a little more care and development effort from Leica, before releasing the M9. Regards, Martin

[jaapv]

Actually, the moire problem is minimal using Capture One as raw processor.

[albireo_double]

Actually, the moire problem is minimal using Capture One as raw processor.

 

I have just tried to process some of my images in C1 and you are right, the moire problem is much less severe there. Looks like I will have to start liking C1 (which I don't as I find the colors in my initial conversions less nice/clean than in LR). I will also try to use the LCC profiles that somebody suggested to me to address the red edge problem.

[jaapv]

I find removing red edge using, for instance, a color replacement brush in CS5 or a desaturating red brush in ACR less complicated. Cornerfix makes a good and easy job of red-edge removal too.

[ho_co]

... constantly reappearing drops of shutter lubricant on the sensor ...

I'm being a bit picky here, albireo, but Leica has said that it isn't shutter lubricant.

 

They don't know what it is, but I think they've made a good case in this regard. Remember, this is the same shutter used in the R8 and R9 (including with DMR), and certainly (in different configuration) also in other brands. It seems to be an M8/M9 specialty.

 

The spots certainly are real; for a while on the forum the idea that it was shutter lubricant was pretty widespread.

 

"Grungy deposits" might be a good way to pussy-foot around the matter, or maybe "shutter-shit." In other threads, several people have suggested other formulations.

[hoppyman]

You are describing a significant problem there and I don't believe that is normal.

 

Perhaps you may like to link to some examples. The forum may be able to offer some comment for you. There has been quite a bit of discussion on the subject across several threads. I don't know if you have seen all of that already?

 

Then you are very lucky since my 28mm Summicron shows the "Italian flag" effect in spades. The guy who sold me the lens says he never experienced the problem on his M9. It drives me crazy. I have read suggestions that the front element may be out of alignment on the lens but I doubt it. Makes the whole proposition of the M9 being a "wide-angle specialist camera" rather dubious, in my mind. I use LR 3.3. as my raw processor.

[charles-k]

Yes, but this observation puts the red-edge problem in the same class as the moire problem the M9 has without doubt - and which is generally accepted as an unavoidable tradeoff for the ultimate image quality.

 

+1

 

I have mentioned before that I have a backup M9 as well, having moved on all my DSLR gear. The red edge isue, is a problem for me on the following lenses:

 

35 Summilux Asph II, 28 Elmarit Asph, 28 Summicron and ZM 25/2.8 Biogon

 

with both M9's and have very different red edge characteristics. Using a generic correction does not work for me.

 

Using C1 Pro, does quickly and accurately removes any red cast with the LCC correction profile. Cornerfix is also an option, should you prefer using LR/CS5. The problem with the red edge, is can be present on any edge and is unique to each sensor. It is not until the correction has been performed, that the red cast is more evident, seeing the before and after effects, particularly when there is any light areas, sky, beach, snow and saturation/contrast is increased in the PP workflow.

[Advertisement]

Hallo,

Look here: Leica M

[Washington]

I read this stuff over and over again and I still do not understand what people

who post this stuff EXPECT.

The freakin' camera is not magic.... it is a well done compromise. Please

read up on it rather than just expecting perfection... It IS NOT perfect. (never will

be)

Now, some really talented fellow designed an application called: Corner Fix!!!

Now: why did he do that???? It's a bit of bother to design software to correct a

problem you know?? Now, all these people are whining about red-edge, magenta edge,

black and blue edge, and God only knows what else.

I am no computer genius by any stretch: but I can follow directions and shoot

a profile for a 12mm Voightlander that works just great.

Honestly, there are times when I really have to wonder why some people bought

this camera in the first place?

If one can't deal with simple post processing... one should not use RAW and

furthermore a point and shoot is the far more appropriate.

[albireo_double]

You are describing a significant problem there and I don't believe that is normal.

 

Perhaps you may like to link to some examples. The forum may be able to offer some comment for you. There has been quite a bit of discussion on the subject across several threads. I don't know if you have seen all of that already?

 

I attach two examples that I already posted on get dpi (my username is baudolino there). I admit that the images were shot in rather extreme lighting conditions that make the phenomenon quite obvious (but I also see it with images shot in nice weather, with any f-stop, and only on the 28mm Cron, never with the 35mm Cron vIV or 50mm Cron). I received some advice on the get dpi forum (of which the LLC profile in C1 looks most interesting to me).

 

My point is not that I am "complaining", I like the M9 and use it all the time but I am, let's say, surprised that there is a combination of the red edge (which I find the most serious problem), the moire (perhaps it is more a software choice issue as C1 is better than LR in suppressing the artefacts, its seems) and the gunk on the sensor. All of these issues can of course be dealt with (a new bottle of eclipse and three boxes of sensor swabs have just arrived, for the "negligible" amount of £137 incl. shipping....).

 

The reason I am surprised is that I used to own an M8.2 and did not experience any of the above and just, perhaps naively, expected the M9 to be the same as my old M8.2, just with a bigger sensor. And no, I did not expect the need to shoot a white reference file with a camera like the M9 (even though I do so with the Sinar back) and yes, I did expect more "perfection" from the M9, based on my experience with the M8! There is no other high-end camera these days (and that includes medium format digital) that has so much discussion about so many "issues" (yes, I know Canon had a problem with AF some time ago, etc.). The above 3 are the issues that I have experienced with the M9 and that I have to address on a daily basis. Fortunately, all of my three lenses focus accurately; I have not experienced any focus shift whatsoever and with the other issues I am prepared to live.

 

I am not sure if Washington's post above brings anything useful to this discussion. I do not feel apologetic about being surprised to find some issues with my M9 and I do appreciate constructive suggestions and advice received from other members. I feel no need to be relegated to "buying a point and shoot", either, if I think images from an M9 and 28mm Cron should not require this level of post-processing. Of course, post processing can be done, but it is a waste of time and should not be necessary, to such an extent, with this camera-lens combination.

 

Alas, having read a lot of discussion on the topic of "red edge" now, I conclude that nobody really knows why it happens, why with some camera bodies/lenses and not others, and there are a few work-arounds that people use and so will I. That's basically it for me, for the moment.

 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!