Red Edge--Which lenses?

[Alnitak]

Advertisement (gone after registration)

I would think that long before the camera is handed over to beta testers, technicians would have to shoot test targets with each lens and carefully measure the results - for evenness, color, resolution, etc. How else could they calibrate the firmware for vignetting and color correction?

 

So it wouldn't be a question of noticing it. But simply a matter of measuring it.

 

You would think, but we have no idea how they did that work, so it may not be as obvious as you would think. That's the problem with speculating about what they did or didn't do, and what they did or didn't know.

 

Jeff

[AlanG]

You would think, but we have no idea how they did that work, so it may not be as obvious as you would think. That's the problem with speculating about what they did or didn't do, and what they did or didn't know.

 

Jeff

 

No we don't know. But don't we have to assume that they must have tested in some way? And either for some reason that I can't explain, the problem didn't show up... or they discovered it and let it slide. This is pretty basic stuff if you can simply shoot a white or grey wall and visually observe it. So it would show up if you measured the file too.

 

There are readily available testing setups for this:

 

DxO Analyzer - Key features

Edited February 13, 2010 by AlanG

[Advertisement]

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

[zlatkob]

Perhaps there was an inkling of warning that such was the market Leica was seeking for the M9 by their selection of "testers" that we know about who were mainly physicians, surgeons and salesmen (not that such people can't be fine photographers, but rather by having more people who take pictures exclusively test the camera, the problem might have been highlighted before the gimicky 9.9.09 release). When one of the few photojournalists on this forum (Noah Addis) uses the camera--guess what? He notices there's a problem.

 

As a general principal, testers should be extremely demanding users. In the camera business, this means full-time highly experienced photographers who know how to push equipment to the edge and over the edge, who take equipment to rough places and use it till it breaks, who know all of the flavors and intricacies of accurate color, who show mastery of black & white, and whose understanding of ergonomics grows out of work under intense pressure. They should be hard-nosed, hardcore tough-as-nails #%@$^&*(!-ers who live and die by the camera, who aren't impressed by brand names, and who aren't inclined to use things that don't work right. Their feedback would be invaluable. If a camera can meet their standards, it can certainly meet the standards of the advanced hobbyist and the collector.

 

And Noah, I gotta say, your signature line has a special irony. In numerous posts about technical issues, HCB's statement about not caring about the technical aspects of photography keeps popping up. In his time, he cared enough to use the best small camera around and to have his work very expertly printed. I dare say, if he were alive and a wide-angle color shooter today, he would care about that red edge too.

[adan]

Alan, I guess my gut feeling is that Leica did essentially the tests your describe, with at least one sample each of the current lenses in the M lineup. But that still leaves some room for surprises:

 

1. In the video with Leica folks on Luminous Landscape, Stephan Daniel specifically mentions that setting the corrections was something of an art, since they wanted to correct for color shifts and vignetting induced by the sensor, but NOT for the natural vignetting or artifacts inherent in the lens (as would show up on film, for example) because those are a part of the character of the lens. And also that they were walking a fine line between over- and undercorrecting, and avoiding too much added noise through brightening the corners.

 

2. It is altogether possible that Leica did write good corrections (maybe even using DxO) - for the specific sample of a lens that they tested. I would NOT assume that such a correction is perfect for every other sample of that same lens model (e.g. 21 f/2.8 ASPH) - there may be variations among the lenses, as built over 10 or 20 years, that were not critical for film use but are critical now. As a off-topic example, the 50mm f/2 lens is actually produced in 5 or 6 slightly different focal lengths - Leica used to put little numbers on the focus scale to indicate whether the lens was a 51.2 or 51.6 or 52mm lens (and adjust the focus cam accordingly). They didn't bother with this for wideangles because the DoF covered the variations - but if there are actually 21.2mm and 20.9mm (and more) "versions" of the 21 ASPH out in the world, who knows how well they work with corrections built using Leica's "tame" house sample of the 21.

 

3. As for older lenses, like my 21mm from 1983 and made in Midland - I might assume that Leica has a sample of that old design available - but maybe not, or maybe it is a 1996 version made in Solms that is just subtly different from mine and produced a perfectly good correction that doesn't translate perfectly to my lens.

 

4. That leaves aside the whole question of it having nothing to do with lenses per se. The amount of - variation - we are seeing reported, sometimes with the same lens or lens model on different cameras, may mean there is a manufacturing inconsistency with either the cameras or the sensors. Again perhaps something totally neglible and within tolerances for most purposes, but that fouls up this one aspect. I don't know Leica's whole test and inspection system - but I suspect they are not testing every production camera with all focal lengths (in all model variations) for red shift (or at least, were not at the beginning of production - maybe they've added that step.)

 

And even so - if they are using the same lenses they originally used to build the profiles, they may be getting grade A number 1 results, but that may not translate into those profiles working well for other units of the lens.

 

I can easily see a scenario where Leica built profiles for edge corrections, and triple-checked them (with their in-house lenses), and saw excellent results. But in the real world of 100,000 other lenses and a production line, something snuck up on them.

 

I'm still trying to figure out how my personal 21 can be red on one side and cyan on the other (I doubt that is a profile Leica would have approved). On my early production M9 (received Sept. 17) but not on the early production demo camera I got to shoot with at the intro a week earlier. Also trying to figure out how a firmware tweak can reduce the red on one side and the cyan on the other simultaneously.

 

You're a Canon user. How did Canon "miss" the AF tracking problems of the 1DMkIII? And why do they need to offer AF "calibration" for some users' specific lenses? I'm sure Canon was and is rigorous in testing cameras and lenses - but something in the real world still managed to sneak up on them.

 

Again, my gut feeling is that this red-edge problem is a very hard nut for Leica to crack - so many "possible" reasons for it, such variable evidence. Is it a production problem, or a profiling problem, or less-than-identical lenses already in the marketplace? Is the solution going to be simply new firmware, or a semi-recall to adjust sensor installations or replace sensors in some cameras? Or will some lenses just need to be sent in for rebuilding to tighter centering specs?

_____________

 

zlatkob - darn right. I've always claimed to be the perfect fool to test whether something is really foolproof!

Edited February 13, 2010 by adan

[ArtZ]

Well, the M8 is out since 2006. It has the red-edge issue as well as the M9. The M8 was used with wide-angle lenses by a lot of people.

Not even marknoton noticed it.

 

Yes, I support your position.

 

Though, once again, the red-edge-problem, we are talking about here is not one of the M9 alone, but as well or bad one for the M8. All M8 users, who make photos with wide angle lenses have lived with it for years without taking notice of it.

 

Uli,

 

Why do you keep saying the red edge issue happens also with the M8 and wide angle lenses? That's not true.

 

If you use the WATE with the Milich adapter, you can get red corners... but if you use it, as I do, with the Leica adapter with the 67 mm UV/IR filter, you won't get red corners (Yes, I know, the WATE + Leica filter holder + a 67 mm UV/IR filter it's not a sexy thing). The problem comes from the filter glass: The 67mm filter has a special formula.

 

But I dont have any red corners with my 21 mm and 24 mm, nor my 28 mm or 35 mm lenses.

 

So, please, leave the M8 alone. The problem on the M9 is a different problem.

 

And I won't buy a M8 until this problem will be sorted out. Period.

[C_R]

The WATE and M9 was tested by Steve Huff in Nov 2009,

 

The Leica Wide Angle Tri-Elmar (WATE) Lens Review | STEVE HUFF PHOTOS

 

and he said there was no red edge issue. I was not convinced, his photos showed the opposite, for example:

with up to 10 points magenta tint in the corners (LAB values). Clearly visible, and distracting for my taste. Sorry for the large photos (linked)

 

So I went for the CV15, and fix the color cast in PP. Not a big issue but annoying of course.

 

The question is, why many other WATE user obviously dont have a problem on M9:

 

is it due do improved firmware, already ?

is it due to sample variation ?

 

When Leica improved firmware algorithms already, they should let us know (maybe, I will drop the §§§ for the WATE...)

[UliWer]

Advertisement (gone after registration)

Uli,

 

Why do you keep saying the red edge issue happens also with the M8 and wide angle lenses? That's not true.

...

 

I keep saying this, because I constantly see red edges on photos I make with the Leica 3.8/18mm, and to a much lessser extent and only in certain circumstances, snow for instance - with the 2/28 and 1.4/35. I have posted many examples - all of them made with the M8 - in this forum since September 2009. Other people have shown examples of the issue made with the M8 as well.

 

As we know from examples of pictures from the M9 the issue is not equally to be noticed with every lens in every situation. The same is true for the M8.

 

I beg your pardon if I bother anybody by saying what I know and see. There is no obligation for anybody to accept what I look upon as evident facts. If I should say, the issue was not occuring with the M8, I were a liar.

[thrice]

You may find, that they calculated the cyan drift rather than measuring it. It would be a lot quicker and a lot cheaper to do it that way, similar to using calculated MTF's and falloff figures instead of measured. An error in this calculation could account for the problems we see, as the calculations would have been used for the M8, but the crop hid most of the problems.

[sandymc]

You may find, that they calculated the cyan drift rather than measuring it.

 

That would be, in my opinion anyway, gross negligence. A $7000 camera with in-camera vignetting correction that was never tested in practice? No way.

 

Given the number of 18mm lenses suffering from this (so far as I can tell, all of them), its simply beyond belief (my belief anyway) that any reasonably competent vignetting test would not have picked this up.

 

I mean, how many people out there have an 18mm that doesn't show some level of red edges on an M9?

 

Sandy

[UliWer]

Here is a thread from 09/09/16 about the issue with the M8 with links to examples I posted in the German Forum:

 

http://www.l-camera-forum.com/leica-forum/leica-m8-forum/98846-red-shift-3-8-18-super.html

 

In #13 of this thread from Sept. 2009 Sean Reid commented this - also mentioning what he knew already about the issue with the M8.

 

More examples with the M8 from other members of this Forum:

 

3.8/18mm:

http://www.l-camera-forum.com/leica-forum/forum-zur-leica-m9/111476-firmware-upgrade-auf-1-00200001-mindestens-5.html#post1179564

 

2.8/21mm

http://www.l-camera-forum.com/leica-forum/forum-zur-leica-m9/111476-firmware-upgrade-auf-1-00200001-mindestens-5.html#post1180145

 

 

My own examples with the M8 and other lenses:

 

35mm:

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!

 

28mm:

 

I hope this all will be an episode from history soon.

[marknorton]

Well, the M8 is out since 2006. It has the red-edge issue as well as the M9. The M8 was used with wide-angle lenses by a lot of people.

Not even marknoton noticed it.

 

It's there on the M8 but to a much less noticeable degree - the M8 neatly crops out the bits we don't really want to see. Put an 18mm Super Elmar on an M9 and there it is, clear as daylight.

 

If Leica did any meaningful testing of the M9, they would have seen it as well; maybe they did such testing but they were so obsessed by the 09/09/09 date that everything else was swept away.

 

I'm sorry, but it simply reinforces my impression that the M9 was rushed to market before it was ready and those of you still waiting for one should continue doing just that.

[UliWer]

Sorry the last picture with the M8 and 2/28 was already corrected by adjustment of white-balance.

 

Here the uncorrected version:

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!

[Hacker]

I prefer testers who own the equipment, not loaned or borrowed from the manufacturer(s).

 

As a general principal, testers should be extremely demanding users. In the camera business, this means full-time highly experienced photographers who know how to push equipment to the edge and over the edge, who take equipment to rough places and use it till it breaks, who know all of the flavors and intricacies of accurate color, who show mastery of black & white, and whose understanding of ergonomics grows out of work under intense pressure. They should be hard-nosed, hardcore tough-as-nails #%@$^&*(!-ers who live and die by the camera, who aren't impressed by brand names, and who aren't inclined to use things that don't work right. Their feedback would be invaluable. If a camera can meet their standards, it can certainly meet the standards of the advanced hobbyist and the collector.

 

And Noah, I gotta say, your signature line has a special irony. In numerous posts about technical issues, HCB's statement about not caring about the technical aspects of photography keeps popping up. In his time, he cared enough to use the best small camera around and to have his work very expertly printed. I dare say, if he were alive and a wide-angle color shooter today, he would care about that red edge too.

[UliWer]

It's there on the M8 but to a much less noticeable degree - the M8 neatly crops out the bits we don't really want to see. ....

 

Does it not seem very improbabable that the red shift was already there with the M8 as we learnt here in this forum that the issue was caused by shifting the M9-sensor?

 

And if you are looking for examples where the red shift on the M8 with the 18mm Super Elmar is clear as daylight you have only to look at some links I posted above.

[D&A]

Hi All,

 

I'll add to somewhat unrelated thoughts to this conversation.

 

1. Quite some time ago with an original M8, there was a particular day where I was shooting a predominantly white facade office building in bright sunlight with a 18mm. It was the first time I noticed quite prominently "red' (magenta) running the entire length of the left side of the frame. It took me by surprise but at times wasn't repeatable to a great extent. It may have been brought on by the combination of subject and strong lighting (and direction of the light). Of course appropriate UVIR filter was on lens and it was coded.

 

2. Without rehashing the possible difficulties of re-writing firmware to address red left edge (with possible right cyan edge)...the overwhelming question for me has been this......... although realizing the issue may be due to multiple factors (particular lens used, sensor, body or lens variations)...if it was due to any number of variances in these items...then why is the red almost always predominantly seen on the left edge (often accompanied by lower bottom of frame)? If there was variation in tolerances of lenses, so that some samples exacerbated the problem due to misalignment, wouldn't it then be random, which edge showed red when various offending misaligned lenses were used? Why always predominantly the right side if the sensor placement is symmetrical?

 

Dave (D&A)

[zlatkob]

I prefer testers who own the equipment, not loaned or borrowed from the manufacturer(s).

Yes! But they're not going to own the equipment before it is released. Testing before release can only be done with loaned equipment.

 

After release, the manufacturer should get ongoing feedback from hardcore heavy-duty users who own the equipment and sometimes break it. Such testers are not likely to be on internet forums much or at all, or writing product reviews. As much as I enjoy and value product reviews, they are written for the consumer, not the manufacturer. A manufacturer needs a different kind of feedback. So I would make a hard distinction between product testers and product reviewers. (I don't rule out that someone can be both, but it's really different role.)

[D&A]

In my post above...the last sentence I wrote questioned as to why always the "right" edge (under the senarios I described). What I meant to type is "why always the left edge", regardless of the various number of possible causes.

 

Dave (D&A)

[AlanG]

...You're a Canon user. How did Canon "miss" the AF tracking problems of the 1DMkIII? And why do they need to offer AF "calibration" for some users' specific lenses? I'm sure Canon was and is rigorous in testing cameras and lenses - but something in the real world still managed to sneak up on them....

 

This entire thread is way out of my expertise so I have no idea what is causing the issue nor do I know the cure. As for Canon, I also have no idea if they knew there was a focus tracking issue and let it slide or simply missed it in testing. I've never used that camera and don't shoot much action where I track. As a matter of fact I have 8 manual focus lenses and only 6 AF lenses. The existence of the 17mm TSE keeps me using this brand regardless of what I might prefer about other systems and formats. I'd probably be the last guy to notice a problem with AF. I'm sure it is hard for Leica to test for everything just as it may be hard for other companies to do so.

 

But it seems to me that in going from the M8 to the M9 this is mostly what they needed to test and calibrate. They may already be close to the limit of how much correction can be made in firmware before the image is too degraded. So perhaps they simply made a compromise. But either they knew or they didn't. And perhaps as you have said, there is significant variation from lens to lens and body to body that can't be cured via a universal setting in firmware. If so, they may need to find a way for fine tuning the firmware by the user or at least a way to do this in Raw processing. (The digital back version of C1 has provision for this.) Keep in mind that I'm totally speculating here.

 

A long time ago, Kodak loaned me an SLRc camera which is a 14 megapixel full frame body that used a Fill Factory sensor. It had pretty extreme color shfts that were "corrected" by selecting the specific lens in camera firmware menus or by using correction tools in the Kodak raw conversion software. When I used shift lenses on it I got very strong color shifts that did not occur when I used the same lenses even at their most extreme shifted positions on the 1Ds. So there was something about the Canon technology that was superior. This was the principal reason I couldn't get behind the Kodak SLRc at the time.

 

It occured to me at that time, that the 1Ds had some kind of unique sensor/microlens/ firmware design that avoided this color shift. I think this may be one reason why it also took Nikon a while to get a full frame sensor from Sony. As they had to come up with a solution that was better than the Kodak work a round that some didn't think worked too well. And it may help to explain that medium format cameras generally require color calibration shots through a plexiglass diffuser when using movements.

 

So what does this have to do with the M9? I think the same thing may be going on although I don't know the specific cause other than to say it may be the sensor/microlens limitation. All one really has to do to find out is to hold a Canon or Nikon shift lens in front of the Leica M9, wrap some dark cloth around the gap and shoot with and without shift. This will show if the M9 sensor is as forgiving as the one in the Canon cameras and Nikon cameras. (I haven't shot with the Nikons so I'm assuming they work OK with shift lenses.) And it would at least tell us that it is not something specific to the Leica lenses but could eventually be cured even if that means a different sensor/microlens will have to be used.

Edited February 13, 2010 by AlanG

[noah_addis]

...They should be hard-nosed, hardcore tough-as-nails #%@$^&*(!-ers who live and die by the camera, who aren't impressed by brand names, and who aren't inclined to use things that don't work right. ...

 

Couldn't agree more.

 

...

 

And Noah, I gotta say, your signature line has a special irony. In numerous posts about technical issues, HCB's statement about not caring about the technical aspects of photography keeps popping up. In his time, he cared enough to use the best small camera around and to have his work very expertly printed. I dare say, if he were alive and a wide-angle color shooter today, he would care about that red edge too.

 

I'm sure he would care. For me the technical aspects of photography only matter when things aren't working as advertised.

 

For some cameras I've worked with--such as the M6 and MP, 501CM, Arca 8x10, F5, D700, technical issues just went away so I could focus on my work. I want my gear to be transparent. It needs to do what I ask it to do with a minimum of fuss and with a great amount of reliability and repeatability.

 

The M8 wasn't quite at that level--the crop factor, filters, etc., were always frustrating.

 

The M9 had a shot, but it's not there yet. We'll have to wait to see if FW can help...

[Advertisement]

Hallo,

Look here: Leica M

[jklotz]

IMHO, the single largest contributing factor is lack of another digital rangefinder camera. I think some competition would be good, and possibly be the motivation Leica would need to make sure problems like this were addressed in a timely manor. Course' I'm a little sour at the moment, having had to send my brand new M9 and 3 lenses back to NJ because the rangefinder is way off, and being told it's going to be 4 weeks. That is after waiting close to 3 months to get the damn thing