New M and old super-wide

[Ecar]

Advertisement (gone after registration)

Well, the CV 12mm doesn't seem to be too weird:

 

http://www.l-camera-forum.com/leica-forum/forum-zur-leica-m9/274421-praxisberichte-und-bewertungen-der-m.html#post2332472

 

Not "too weird" - but still rather nasty, I'd say...

[mjh]

Time for supplying some theoretical background it seems …

 

There are two different issues we are dealing with here. One is the variation of the incident angles between (chief) rays from the center to the corner of the image. With near telecentric lenses there is little variation and the incident angles stay small, but especially with wide angle lenses, rays hitting the corners of the sensor are much greater than those hitting the center. So the second issue is that this variation depends on the lens.

 

The traditional way of dealing with the first issue is microlens shifting. One takes an ideal or typical lens with a certain variation of incident angles and devises a microlens pattern optimised for that lens. Obviously this has to be a compromise; one chooses a microlens pattern giving optimum results for the lenses used most often and still acceptable results with the more extreme kinds of lenses. Software corrections take care of whatever there is left to correct (hopefully).

 

Now the new M was designed with not just M, but also R lenses in mind. When the M9 had issues with very short focal lengths and a steep increase of the incident angle towards the corners, R lenses with long focal lengths would exacerbate the issues at the other extreme, namely small incident angles even in the corners. There is just no way to deal with this using microlens shifting since any change improving results with one type of lenses would worsen the performance with other kinds of lenses. So the idea was to design microlenses accepting a wide range of incident angles, rendering microlens shifting unnecessary. Thus the cucumber-shaped microlenses of the MAX sensor that are supposed to refract rays so they would always hit the photosite underneath, regardless (mostly) of the incident angle.

 

Ideally this would solve both issues at once, whereas the M9 only solves one, and not fully at that. Maybe it doesn’t, at least not without a little help by firmware-based corrections, but then it is a much greater challenge the M tries to meet here.

[Peter H]

Michael, does this suggest that a "better" sensor solution could have been found if there had been no attempt to accommodate R lenses in the new camera and the developers had concentrated all efforts on the best possible performance for M lenses?

[mjh]

Michael, does this suggest that a "better" sensor solution could have been found if there had been no attempt to accommodate R lenses in the new camera and the developers had concentrated all efforts on the best possible performance for M lenses?

Impossible to say for certain right now, but I doubt it. The alternative to Leicas approach would have been to stick with microlens shifting M9-style. Results might have been slightly better than with the M9, due to the smaller gap between microlenses and photosites, but not radically so.

[Peter H]

Thanks Michael. Your explanations are very clear and helpful, and reassuring too!

[douglasf13]

Did a quick and highly unscientific test with the CV 21/4 against the usual white wall trying various codes.

The one that seems to work best is that for the WATE.

Still quite a bit of red edge, but much better than with the first iteration of the M9 firmware.

Whilst we shouldn't expect Leica to come up with a solution for 3rd-party lenses, perhaps a future FW will improve things further.

Once I'm back in front of my computer, I will also check the effects of the lens settings in LR 4.4 RC.

And there's always CornerFix.

 

If you're using LR4, you can use Adobe's flat field converter, rather than CornerFix, and it'll be a lot less to add to your workflow. It works great.

[jonoslack]

Advertisement (gone after registration)

HI There

There were two things which were fairly clear when testing the cameras:

 

1. there was much less of a problem with colour shift than with the M9 - even coding the lenses quite wrongly generally caused little difficulty.

2. There was a possibility of red vignetting with some lenses which were not covered by corrections.

 

Personally - I had zero problems with any of my lenses:

WATE

21 pre asph elmarit

28 'cron (although some others did have red edges)

35 summarit

35 FLE

 

I did try a 15CV, and it produced pronounced red vignetting (again, with very little colour variation). Other CV lenses I tried seemed to do quite well, but I don't have access to many 3rd party lenses.

 

My instinct in the versions of the firmware I was using was that there was no vignetting correction (because coding lenses quite wrongly didn't seem to have an effect) - and that it was still pretty good.

 

I understand that the final release software has sorted the issue with the 28 'cron completely, but it seems that the 21 pre Asph elmarit is now quite bad . . . . . . . but that coding it as a 28 elmarit asph largely sorts it out. (as far as I understand it, that 28 elmarit asph needs the strongest correction of any lenses . . . . could be wrong though).

 

My take on all this - is that there is likely to be an issue with some older lenses for a few weeks until they're all fixed in firmware, but that there is likely to be an easy workaround (see above).

 

The good news is that the problem seems to be largely red vignetting, and its pretty much symmetrical (unlike the weird and wonderful colours the M9 could produce) Which should mean that it's pretty straightforward to sort it out.

 

Whether it's possible to get perfect results with the CV15 is a moot point (the CV12 should be okay though).

 

I hope this helps - I really don't think it's time for panicking!

 

O1af - I'd be fascinated to hear how your rokkor works when it's coded as a 28mm elmarit asph.

 

all the best

[chris_tribble]

I've played with the 21 Elmarit pre-asph. Using the current profile, it's not good. Using Jono's idea of the 28 Elmarit 2.8 it's improved. I think Leica can and should be able to improve the situation with the 21. I've not tried my 18 Zeiss - but expecting Leica to fix things for non-leica lenses isn't reasonable! FWIW I'm on FW 1.102.

 

1. is coded as 21

2. is coded as 28

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!

[UliWer]

The same lens sometimes showing strong colour shift (chris_tribble's example with the 21mm Elmarit pre-asph.) and sometimes none at all (jonoslack's experience with this lens) does not mean that one observer is wrong. This was the same with the M9 and my theory about it was that it depends on the light. With diffused lighting (the blinds in front of the window of chris' example...) the risks of getting the red edges are much higher than with clear light where a polarizing filter might have its maximum effect.

 

To make up one's mind whether there are red shifts in the corners one should try a photo with an opal plane right in front of the lens. Thus eliminating different light conditions you get rather reliable results.

[douglasf13]

To add a bit to my post above, I don't think color shift is really a big deal anymore, unless you shoot jpeg and don't do any post processing. Both Lightroom 4 and C1 have the option of fixing color shift in the converter, and it only takes a couple more button clicks than my usual workflow.

 

In fact, I've begun turning of lens correction altogether in camera, because it is the vignetting correction that bothers me, especially in lowlight, where the edges of the frame don't look that great, because of the exposure boost for the areas with vignetting.

 

The good news to me is that it looks like there is less vignetting with the new M sensor.

[algrove]

@ Jono or Chris

 

Did either of you try a wide R lens on the M? I recall using the 28-90 at Photokina with no apparent color shift, but then again that venue was difficult to see any shift anyway. I believe it was you Jono who published one image taken with the 28-90 R. No?

 

Hey, if the WATE works well then that is encouraging.

 

I will be trying a 15/2.8 R with a Novoflex adapter. I see that lens is listed as being supported in the M brochure (English page 20). I see they also list the 19/2.8 which I will try too, thanks to a friend who has both.

[jonoslack]

@ Jono or Chris

 

Did either of you try a wide R lens on the M? I recall using the 28-90 at Photokina with no apparent color shift, but then again that venue was difficult to see any shift anyway. I believe it was you Jono who published one image taken with the 28-90 R. No?

 

Hey, if the WATE works well then that is encouraging.

 

I will be trying a 15/2.8 R with a Novoflex adapter. I see that lens is listed as being supported in the M brochure (English page 20). I see they also list the 19/2.8 which I will try too, thanks to a friend who has both.

 

HI There

I've used the 28-90 extensively, and the 21-35 less so. They're all absolutely fine (but you'd expect them to be okay, as the angle of light is much less severe than on M lenses).

 

all the best

[plewislambert]

Very interesting because I was until recently using an M4-P film camera with Voigtlander 15mm and 21mm ultrawides which have the rear glass close to the film plane. The results were good but when I repeat the process on my M9 the results are vignetted and discoloured at the far ends of the frame, which I treat with Cornerfix, a free program. The underexposed edges are corrected by this program and the colour improves but the digital noise can be very obvious at the underexposed edges, less so when I give 1 stop extra exposure than the meter suggests. This can lead to blown highlights (e.g. the windows of a church interior shot) but some detail can be recovered from the DNG image.

To save all the bother I bought a Tokina manual 17mm slr lens that I use with a Leica M adapter and it gives acceptable results, probably less contrasty than the Voigtlander 15mm. If I get some sunshine in London I can take some more interiors and see if I can improve the contrast and sharpness.

A supplementary viewfinder is essential- I am still experimenting.

Philip.

[algrove]

HI There

I've used the 28-90 extensively, and the 21-35 less so. They're all absolutely fine (but you'd expect them to be okay, as the angle of light is much less severe than on M lenses).

 

all the best

 

Yes , they're Leica lenses (thank gosh), but it seems the R glass is the way to go with both R wide and with R telephotos which allows one to use MUCH longer telephotos on the M than anything Leica has for the M today and maybe into the foreseeable near future.

[jonoslack]

Yes , they're Leica lenses (thank gosh), but it seems the R glass is the way to go with both R wide and with R telephotos which allows one to use MUCH longer telephotos on the M than anything Leica has for the M today and maybe into the foreseeable near future.

 

Had a lot of fun out shooting with the 80-200 f4 today - under-rated lens, and manageable size too - snap 'em up whilst they're still cheap!

 

all the best

[Hannes Lummes]

What is the focal length at which M9 was most neutral? And how about M? (assuming fairly symmetrical lens design)

 

This would be very nice to know. I would test it right away...if I had those cameras.

 

I must say that I'm a bit disappointed that M sensor couldn't quite solve the incident angle issues yet. Perhaps a real revolution in sensor design is needed before we see that happen.

 

The idea that R lenses would be any kind of solution to super wide is absurd. They don't get even near super wides designed for mirrorless systems, the *real* advantage of range finder optics.

[adan]

It's probably worth repeating (since it has been three years since our M9 (red edge) discussions began), that when it comes to interactions between wide lenses and digital sensors:

 

Not all lenses are created equal - even if they have the same focal length or field of view. It is the path the light rays follow through the glass and exiting the glass that count. How "wide" the lens is, or how close it is to the sensor - in and of themselves - don't matter. Although that will have an indirect influence on the light paths, obviously.

 

The ultra-short 12mm C/V produces less color distortion than the 15mm - because it bends the light rays differently and happens to do so in a slightly more "sensor-friendly" way. (Both were designed in the film era).

 

The Leica WATE at 16mm or 21mm produces less red-edge than the Leica 21 pre-ASPH, and far less than 01af's Rokkor or the Leitz Super-Angulon 21s.

 

A 28mm lens for an SLR is designed to sit 40-45mm from the image sensor/film (to leave room for a swinging mirror), and thus bends the light very differently from a 2006 28 Elmarit ASPH, or a 1964 28 Elmarit.

 

The Digilux 2's lens had a focal length as short as 7mm, and the back element virtually touched the sensor (see attached), yet produced no color shifts at all. Partly because it was designed for digital in the first place, and that back element was in part a "corrector plate" - in effect, one huge "microlens" - that redirected the light rays to a sensor-friendly angle of incidence.

 

Note how the "red" edge rays are redirected to be as "vertical" as the green center rays.

 

All of which is just to say that you have to know how a given optic is designed - not just the focal length engraved on the front.

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!

[chris_tribble]

Had a lot of fun out shooting with the 80-200 f4 today - under-rated lens, and manageable size too - snap 'em up whilst they're still cheap!

 

all the best

 

"cheap"? Maybe "not too outrageously expensive" might be more accurate

 

Certainly I'm very glad I've got mine...

[UliWer]

To add a bit to my post above, I don't think color shift is really a big deal anymore, unless you shoot jpeg and don't do any post processing. Both Lightroom 4 and C1 have the option of fixing color shift in the converter, and it only takes a couple more button clicks than my usual workflow.

 

...

Well...

 

.... (please don't misunderstand the following, and perhaps someone will believe me that I am not too fond of Leica-bashing, but....)

 

you pay 6.200,-€ for a camera + some thousands for some lens, only to concede, that it's not really a big deal anymore, if your photos are spoiled by red edeges on the right side of the photo - when it was more on the left side with the M9?

 

You need software to manipulate the results of the most expensive lenses you can buy for money (and some other lenses as well) to make them usable?

 

I beg Leica's pardon, but I say no to this attitude.

 

We had the experience since the M8, more obvious for the M9 and customers stayed patient. There was a firmware remedy for the M9, which I think worked, and I do not complain about the red edges of the M9 - anymore (about one and a half years since it came out in 2009 there was something to complain about).

 

Jenoptics was responsible for the electronics of the M9, not Leica - though customers paid Leica, not Jenoptics. The sensor was bought from Kodak though costumers paid Leica, not Kodak.

 

Now the sensor and the electronics are made by Leica. They knew what they did and they knew about the problem. I remember precisely what was said in certain interviews, when the M 240 was announced at last Photokina: the colour-shift problem does not exist anymore. Since September 2012 there was some time to cope with it.

 

Now we see: the problem is still there.

 

Today I can only say: Leica failed. If there is no technical solution for this problem, they failed to tell the truth.

[Advertisement]

Hallo,

Look here: Leica M

[k-hawinkler]

"cheap"? Maybe "not too outrageously expensive" might be more accurate

 

Certainly I'm very glad I've got mine...

 

 

Hi Chris,

 

$1600 to $1800 USD seems to be the going rate currently on Ebay.

I am not good at bidding, so the pay-it-now-price it typically is for me unfortunately.

I am glad I got my 80-200/4 though. It has become a favorite lens on my NEX-7.

Still fairly easy to focus and shoot handheld.

Certainly pretty costly. But not yet in the stratosphere as other R lenses seem to be.