Jump to content

Corner Color Drift on M10?


cookedart

Recommended Posts

Offset microlenses would have a detrimental effect on more telecentric designs.

This is an interesting sentence.  :D

Please explain what is the meaning in this context. E.g. what does it mean for the "newer" M lens designs. (WATE, 1.4/28, Apo50)

Edited by steppenw0lf
Link to post
Share on other sites

Do we have this "diagram" also for the Q and SL sensor ? Automatically many have assumed that the same principle is used there as well (at least for the SL). But I have never seen an official documentation about this. (Only for the Ms ).

So can we safely assume it's identical ? Or very similar ?  (You did in your note, too. @jaapv) (So do I).

Q and SL sensor are (probably) from TowerJazz, M240 (and M10 probably) from CMOSIS. So would this (two different sources/designers) implicate "differently" shaped microlenses ?    

Edited by steppenw0lf
Link to post
Share on other sites

Interesting that there are third-parties offering a replacement for the Sony camera.

But if I was a user I had other priorities and would mainly be worried, if the native (Sony/Zeiss) lenses are still working as well as before the exchange. I would never do this "operation" to my camera.  :D

 

That is one of the advantages of the SL - I am glad that Leica handled it for me and that no "special treatment" is necessary.

Edited by steppenw0lf
Link to post
Share on other sites

Advertisement (gone after registration)

This is an interesting sentence. :D

Please explain what is the meaning in this context. E.g. what does it mean for the "newer" M lens designs. (WATE, 1.4/28, Apo50)

Like you said, the SL seems to perform better with the 28 lux. Leica has to compromise in some way since the optical design of m lenses varies, but certainly a lot of the lenses have light rays with steep light angles. Someone would have to test these newer lenses on the m9 to see what effect it has. My guess would be the m9 performs better with older lenses (like the 28 cron for instance) but worse with the 28 lux vs the m240 and SL.

Link to post
Share on other sites

Like you said, the SL seems to perform better with the 28 lux.

The A7 Kolari mod actually shoots that lens as well as M240, according to Digilloyd and Kasson. So strong SL performance makes sense.

 

But, is the SL better than the 240 with it?

 

I have a hard time believing that, but I'm always learning new stuff :)

 

My contention is: anything great with M glass will be great with R glass. The reverse may not be the case.

 

I admire SL camera very much. Very versatile, good ISO, good files, super build, etc. Way nicer than Sony in so many ways, but too big for me. Right now I use these two:

32277225612_5a440a5410_c.jpg

DSC09513 by unoh7, M9/A7.mod

 

But I need better performance with more M wides in my next Sony, good as the Kolari v2 is, here with 28 cron:

 

17271630986_64297b0c5c_b.jpg

Rapunzel Lived Here by unoh7, f5/6

 

No color shift with this lens, but a little FC left at F/2. Looks like it was taken on digital M, no?

 

17109818228_711c2cb5c2_b.jpg

Welcome to Shoshone by unoh7, f/8

 

That lens is unusable on stock A7. No profiles applied, or cornerfix/flat field. Whole point is: coverglass and filter stack are biggest factors comparing Sony A7 and Leica sensors with M glass. SL specs on that front are unclear. I would love to know them. Someday we will.

 

Sony does claim off-set micro lenses. They have four distinct sensors: A7/A7ii, A7s/A7sii, A7r, A7rii (BSI) Even stock they all have subtle differences with M glass.

Edited by uhoh7
Link to post
Share on other sites

My contention is: anything great with M glass will be great with R glass. The reverse may not be the case.

 

...

 

Sony does claim off-set micro lenses. They have four distinct sensors: A7/A7ii, A7s/A7sii, A7r, A7rii (BSI) Even stock they all have subtle differences with M glass.

This is what I'm not really sure about. Not many m9 users shot r glass for instance. I hadn't heard of a7 bodies using offset microlenses. In fact the only sensor I heard of using it was the Kodak sensors in M bodies.

 

Anyways I managed to get an m10 body, so I'll be sure to test both the Zeiss 15mm ZM and the voigtlander 15mm on it if it comes tomorrow as planned. Both lenses have red corners no matter what profile.

Edited by cookedart
  • Like 1
Link to post
Share on other sites

I think a few clarifications are in order …

 

The issue at hand is caused by large incident angles of the light hitting the sensor. This is a general issue and the standard solution originally employed by all camera manufacturers was microlens shifting. This assumes a certain behaviour of the lens, i.e. one assumes that the incident angle grows from 0 at the center of the image toward some larger value in the corners. Each microlens is then shifted from its position directly overhead its sensor pixel so the incident light (at an angle dependent on the position of the pixel) will still hit the centre of the pixel.

 

This works to some extent but it is a compromise solution. You can optimise microlens shifting for some troublesome wide-angle lenses, but then the sensor will not perform as well with normal or telephoto lenses. One has to strike some compromise and thus there will be some lenses with less than optimum results.

 

With the M (Typ 240) Leica switched to a different approach. The requirement of also supporting R lenses was one reason, but that was only exacerbating the issue that microlens shifting would only ever yield optimum results with one kind of lens. The idea was that the sensor should be less dependent on incident angles in the first place, and that was achived by using microlenses with a shorter focal length and by reducing the space between microlenses and the sensor chip (the latter required some changes to the design of a CMOS sensor). Since then, this approach has been adopted by some other vendors as well – for example, Fuji does the same.

 

Leica wouldn’t go into the details of what approach they have chosen with the M10 but the main point is that the M10 shares the characteristics of the M (Typ 240) in that it is largely independent (and actually even less dependent) of incident angles.

Edited by mjh
  • Like 7
Link to post
Share on other sites

I think a few clarifications are in order …

 

The issue at hand is caused by large incident angles of the light hitting the sensor. This is a general issue and the standard solution originally employed by all camera manufacturers was microlens shifting. This assumes a certain behaviour of the lens, i.e. one assumes that the incident angle grows from 0 at the center of the image toward some larger value in the corners. Each microlens is then shifted from its position directly overhead its sensor pixel so the incident light (at an angle dependent on the position of the pixel) will still hit the centre of the pixel.

 

This works to some extent but it is a compromise solution. You can optimise microlens shifting for some troublesome wide-angle lenses, but then the sensor will not perform as well with normal or telephoto lenses. One has to strike some compromise and thus there will be some lenses with less than optimum results.

 

With the M (Typ 240) Leica switched to a different approach. The requirement of also supporting R lenses was one reason, but that was only exacerbating the issue that microlens shifting would only ever yield optimum results with one kind of lens. The idea was that the sensor should be less dependent on incident angles in the first place, and that was achived by using microlenses with a shorter focal length and by reducing the space between microlenses and the sensor chip (the latter required some changes to the design of a CMOS sensor). Since then, this approach has been adopted by some other vendors as well – for example, Fuji does the same.

 

Leica wouldn’t go into the details of what approach they have chosen with the M10 but the main point is that the M10 shares the characteristics of the M (Typ 240) in that it is largely independent (and actually even less dependent) of incident angles.

You imply, and it may well be the case, that the M9 CCD is somehow more optimized than 240 for M wides, or do I read this wrong?

 

Regarding Sony:

When the original A7r came out it was stated to have:

"Gapless, optimally positioned on-chip lenses" and supposedly

"Sony optimized the design and positioning of every on-chip lens (OCL) covering every pixel to significantly enhance light- gathering efficiency. A gapless on-chip lens design eliminates the gaps between the micro-lenses to collect more light. Moreover, each on-chip lens is optimally positioned depending on its location to accommodate the sharper angle of light entering the periphery, which is caused by larger sensor dimensions being teamed with the E-mount’s short flange-back distance."

 

Now that is funny because the A7r is the worst of all with M wides, showing more colorshift than any other model. I tested it extensively in December 2013. I sent it back after doing so and got a plain A7. It was still inducing FC but did not shift color near as much. But it's poor performance with my zm18/4 and 28 cron and 35 skopar caused me to buy a used M9. :) Thank You Sony!

I did not use the A7 much for the first year I owned the M9, just for some longer lenses. Then a A7 user contacted Kolari Vision hoping to get it working with the 21/4 skopar. Astro people had already been working on the SOny filter stacks. Kolari picked a IR cut like the M9 and threw out the rest of the filter stack, then moved the sensor forward. I sent mine in. This was fortuitous because a week after it came back I lost the shutter in the M9 and it spent two months in New Jersey. The Kolari jumped right in, though of course it could not do the 21 Skopar. The 35 was OK though. :)

 

It has been an awesome 2nd back for M, and great for SLR glass too.

 

Anyway, I believe the Sony sensors would be equal to any M sensor if the clear .7mm coverglass was replaced by Schott S8612 or something like it, at .8mm, the filter stack removed, and the sensor moved forward for infinity. This is entirely possible to do, but coverglass replacement is highly specialized, and the normal sensor techs will not do it. I do know places that do it all the time with many sensors, but nobody asked them yet with a Sony. 

 

The link I gave previously shows A7rii with NO coverglass, and stack replaced by .5mm IR cut in the stack, and sealed by a tech in Taiwan. Kolari is experimenting with new ultra-thin IR cut filters now. We'll see how that goes. I will mod a A7r2, with it's incredible BSI sensor, for M pretty soon, once I can see all the possibilities and pick one. 

 

That will replace my current A7 Kolari as second back to M9, in a very complimentary way: I'll get ISO 6400 (like M10), 42MP, 4K video, Silent Shutter, and with the Techart Pro adapter I'll also get excellent auto focus with any M lens. While I don't often care, for events this can be very useful. The r2 will also shoot my SLR glass a bit easier than M10. But who knows, maybe I'll be able to replace M9 with M10 as well, some day. For sure I want a real M as one of my bodies, as they are so nice to use. :)

 

I admire the M10 very much :)

Edited by uhoh7
Link to post
Share on other sites

You imply, and it may well be the case, that the M9 CCD is somehow more optimized than 240 for M wides, or do I read this wrong?

The M9 is (slightly) optimised for wide-angles, but the M (Typ 240) – and the M10 – is optimised for a wide range of lenses rather than just specific lenses.

 

Regarding Sony:

When the original A7r came out it was stated to have:

"Gapless, optimally positioned on-chip lenses" and supposedly

"Sony optimized the design and positioning of every on-chip lens (OCL) covering every pixel to significantly enhance light- gathering efficiency. A gapless on-chip lens design eliminates the gaps between the micro-lenses to collect more light. Moreover, each on-chip lens is optimally positioned depending on its location to accommodate the sharper angle of light entering the periphery, which is caused by larger sensor dimensions being teamed with the E-mount’s short flange-back distance."

The last sentence refers to microlens shifting. A gapless microlens design means there aren’t any gaps between microlenses, i.e. there is no light lost in the cracks. But as far as reducing vignetting and colour-shift is concerned, it is key to reduce the gap between the microlens layer and the sensor chip – quite a different issue.

  • Like 1
Link to post
Share on other sites

The M9 is (slightly) optimised for wide-angles, but the M (Typ 240) – and the M10 – is optimised for a wide range of lenses rather than just specific lenses.

That may have been the intention. In my experience, it is not the case...but if somebody has some evidence:

 

27873321870_037c9277da_b.jpg

Top of the Salmon by unoh7, M9 APO 135/3.4 F/5.6

 

30614684421_5469a4de5b_b.jpg

Dauphin by unoh7, M9 135/2.8 Elmarit POF evident in fabrics.

 

These lenses are also very good on the thin-filter 24mp Sony. Better? Not that I have seen. But somebody may have better eyes:

 

30588616241_4b8cf8960e_b.jpg

Fall Banks by unoh7, 135 APO A7.mod F/5.6

 

22238923563_0537727e20_b.jpg

Old Jeepster by unoh7, WO 135 APO on A7.mod

 

23117337601_445ace816a_b.jpg

Storm Cellar by unoh7, Elmarit 135/2.8 WO on A7.mod

 

Elmarit is my favorite 135 :) Might new the most under-used, underrated M Lens. I shoot it often on M9 indoors and out. Seems about equal on both cameras....The M9 is for sure better with some wides, but again the glass over M9 sensor is half the thickness of glass over mod sensor. If you see how that effects the Sony sensor in the filter stack, from 1.9mm to .8mm to an extreme degree with even 35mm M wides, you would guess that is accounting for it.

 

Note the edges on this A7r2 with only .5mm total glass over sensor, zeiss M hologon 15mm, no profile:

https://flic.kr/p/NuXKGh

That also shifts colors on M9. But the edges seem fantastic on the Sony , no? Only 2 or 3 A7rii bodies have been modified to this degree, LESS glass over sensor than M9! Unprofiled, I don't think any M will beat it with wides on the edges. I could be wrong.

 

Look what he is testing next:

https://flic.kr/p/Qcrjzj

That will be very telling...one of the great film wides which has never been good on any digital.

Edited by uhoh7
Link to post
Share on other sites

The M9 is (slightly) optimised for wide-angles .....

 

My experience of the M9 and wides is one of some degree of variability. My guess being that even individual lenses vary to some extent, though why this should be puzzles me. Also, whilst I find that the 21SEM can deliver excellent, even images with no obvious colour drift. Excessive adjustments will often show up some drifting though so any imperfections will out. The M9 is as you say a compromise although if I'm totally honest, its one which is quite acceptable - indeed it works surprisingly well with all the lenses I currently own from 21 to 135 except of course the old 3.4 Super-Agulon :( (now relugated to being a B&W lens - which its good at :) ). Having spent time shooting with the 135/4 (E46) yesterday I have to say that it produces very nice images indeed stopped down, which is fortunately where I tend to use it. A shot is attached which should show any problems if there were any but does not:

 

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!

Edited by pgk
  • Like 3
Link to post
Share on other sites

Charlie, I think up to 135mm, the offset microlenses do not show any discernible image issues in the corners. It's probably the longer telephotos which one might use on LV cameras that would start to suffer in the corners as the perpendicular light rays will be only partially visible to the offset microlenses hence the microlens design of the M9 was not feasible for the M bodies that are LV capable.

  • Like 2
Link to post
Share on other sites

On 09/09/09, one of my very first experiments with a demo M9 was to test how the 24x36 sensor and microlenses worked with a 135. Since I had concerns similar to cookedart's.

 

Used the 135 with and without a profile, shooting simply a section of pure blue sky to look for any corner darkening if the lens was unrecognized. There was none (unfortunately i tossed those images long ago, since they showed no difference - apologies).

 

And the 135 is a pretty darned "telecentric" and perpendicular-ray lens. The light rays are already so parallel that the 135 casts very sharp shadows of sensor dust even at f/4, where a wider lens with technically more DoF (e.g. 35mm) won't reveal the same dust as starkly until f/8 or smaller.

__________________

 

Off-topic for this thread, but on-topic for the M10 - the 135 has really become my favorite M telephoto. I could shoot forever with just a 21/35/135 combo - except for the trickiness of focus @ f/4, and the lack of a large aperture** in dim light, since it requires a shutter speed of 1/250 or higher to prevent visible camera shake.

 

Which is a major part of my decision to jump on the M10 - higher-mag finder for better focus precision and a slightly larger framing area, and the (at least) 3-stop usable ISO increase over the M9, which will make my Tele-Elmar f/4 the equivalent of a 135 f/1.4 (!!) on the M9 in dim light. Boy, have I dreamed of this day....

 

If 21/35/135 seems an unlikely working set, I have an M6 brochure from 1984 in which Leitz/Leica themselves suggest lens combinations/working sets in the back. Including "the wide bridge across focal lengths" set of 21 Elmarit/50 Summilux/135 Tele-Elmar.

_____________

 

***yeah, I know there is the goggled 135 f/2.8. Nice lens - but frankly, if I was willing to haul around gear that large, heavy, and ungainly (and pay the chiropractor bills), I'd still use SLRs.

  • Like 3
Link to post
Share on other sites

Some quick anecdotal tests of the m10 that just arrived:

 

- I still see some red edges on the Voigtlander 15mm v2. One thing to to note - I've manually coded it, and the menu is greyed out for me to manually select a lens code. I believe I have it coded as a WATE. I may remove the coding so I can test it with other profiles (like the 21mm 2.8 elmarit). I believe this is a new behaviour though (to not allow manually overriding the lens coding).

 

- I do not see any red drift with the Zeiss 15mm ZM distagon, which is a definite improvement over the M typ 240. 

 

Please note these are not scientific tests. I will have to do those later.

Edited by cookedart
Link to post
Share on other sites

Edward, I'm betting you will love M10 :) Can't wait to see your style, which I have enjoyed for years now, through the M10 prism. :)

 

On 09/09/09, one of my very first experiments with a demo M9 was to test how the 24x36 sensor and microlenses worked with a 135. Since I had concerns similar to cookedart's.

 

Used the 135 with and without a profile, shooting simply a section of pure blue sky to look for any corner darkening if the lens was unrecognized. There was none (unfortunately i tossed those images long ago, since they showed no difference - apologies).

 

And the 135 is a pretty darned "telecentric" and perpendicular-ray lens. The light rays are already so parallel that the 135 casts very sharp shadows of sensor dust even at f/4, where a wider lens with technically more DoF (e.g. 35mm) won't reveal the same dust as starkly until f/8 or smaller.

__________________

 

Off-topic for this thread, but on-topic for the M10 - the 135 has really become my favorite M telephoto. I could shoot forever with just a 21/35/135 combo - except for the trickiness of focus @ f/4, and the lack of a large aperture** in dim light, since it requires a shutter speed of 1/250 or higher to prevent visible camera shake.

 

Which is a major part of my decision to jump on the M10 - higher-mag finder for better focus precision and a slightly larger framing area, and the (at least) 3-stop usable ISO increase over the M9, which will make my Tele-Elmar f/4 the equivalent of a 135 f/1.4 (!!) on the M9 in dim light. Boy, have I dreamed of this day....

 

If 21/35/135 seems an unlikely working set, I have an M6 brochure from 1984 in which Leitz/Leica themselves suggest lens combinations/working sets in the back. Including "the wide bridge across focal lengths" set of 21 Elmarit/50 Summilux/135 Tele-Elmar.

_____________

 

***yeah, I know there is the goggled 135 f/2.8. Nice lens - but frankly, if I was willing to haul around gear that large, heavy, and ungainly (and pay the chiropractor bills), I'd still use SLRs.

haha watch it, that's my favy ;) It is an ugly hunk. I have 10+ 135mm lenses. None I prefer to 135/2.8 Elmarit, and that has been a surprise to me. That extra weight is not for nothing. Focus accuracy is dramatically better.

 

30614688421_dfbe02031a_b.jpg

Take My Hand by unoh7, on Flickr

 

Look at this hit WO:

30952602056_99c3b8e560_b.jpg

Crux by unoh7, on Flickr

You can shoot it WO and hit nearly everything in spite of tiny DOF, and with this speed M9 is good at indoor events.

 

30988000865_1ed7911890_b.jpg

At River Run by unoh7, This is Mandler!

 

So when I am in any contained area I use elmarit, even on short walks. But when I need to be light, it's the APO. :)

 

But, my defense of the ugly duckling/turkey aside, you make a great point. RF is more accurate in M10. That is good. :)

Edited by uhoh7
  • Like 2
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...