A Sane Attitude to Rededge

[adan]

Advertisement (gone after registration)

Alan, just estimating how much you are tilting an SLR lens is not persuasive. The angle of incidence AT THE SENSOR is what counts.

 

The long register distance reduces the basic angle of incidence for a Canon SLR 17mm substantially compared to a Leica RF 18mm. I.E. the Canon lens, sitting out on the front of a mirror box may have an angle of incidence (not angle of view - different things) for the corners of the frame of 20° (without tilt or shift), while the Leica 18mm is probably more like 35°. Lots of leeway for tilt before even getting close to the challenges the Leica sensor has to handle.

 

If someone has a real knowledge of the location of the exit pupil, nodal point, or whatever it is that is the critical spot from which to measure AoI for both lenses, they can no doubt calculate the precise angles involved and do a more exact diagram.

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!

[Advertisement]

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

[jaapv]

If that is a lens that works on a DSLR without requiring mirror lockup, then it is nice and telecentric, not likely to cause red edges.

 

scott

Exactly my point.

[AlanG]

I looked very carefully at my 17mm lens and while of course it has a retrofocus design, it is hardly what one would call "telecentric." Viewing from the rear, the lens can be swung over quite a wide angle... which is necessary for the wide image circle in order to accommodate the shift and tilt.

 

Consider that there are differences in chip design. The rap on the MF Kodak chips is that they have deeper wells than the Dalsa chips have. This leads to more color shifting. The same problem has been in other chips. (E.g. the Kodak DCS full frame cameras.)

 

Explain why the CV 15mm lens seems to work ok in the 5D report that I linked while it has issues with the Leica chip.

 

"Conclusion

 

So, we can pluck a rangefinder lens out of a Leica; we can adapt the mount and, with a little trepidation, we can sweep aside the mirror and force it down the gullet of our Canon digital cameras. But should we?

 

The Voigtlander 15mm Super Wide Heliar is sharper across the frame at pretty much all apertures. It's corner performance at f5.6 and f8 is much better than the Sigma 12-24mm. It is certainly amazingly compact and portable on the camera, and makes a fabulous focus and forget hip shooter out of your 5D, with superb wide aperture performance... but every shot requires vignetting correction, the viewfinder isn't the last word in utility and you do have to go the extra mile in forgoing even aperture priority metering."

 

He didn't say anything about there being a color shift.

 

Here is my read on what is going on with a 17mm tilt/shift lens. I think trying one of these on a Leica M9 would help answer the question. And since the 5DII has live view (sort of a mirror lockup), it would be pretty easy to adapt a Nikon F mount CV 15 to try it on that model.

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!

[mjh]

Consider that there are differences in chip design. The rap on the MF Kodak chips is that they have deeper wells than the Dalsa chips have. This leads to more color shifting.

Speculation, conjecture, or observation? I am not aware of color shift issues with Kodak’s MF sensors and I don’t see why the design of Kodak’s MF sensors should lead to color shift.

 

The same problem has been in other chips. (E.g. the Kodak DCS full frame cameras.)

Kodak’s full-frame DSLRs didn’t even use Kodak CCDs but CMOS sensors designed by Fill Factory in Belgium.

 

Here is my read on what is going on with a 17mm tilt/shift lens. I think trying one of these on a Leica M9 would help answer the question.

Hasn’t Jaap already answered the question? The relevant difference between a DSLR and the M9 is the flange distance. Adapting a T/S lens for a Canon EOS to the M9 would demonstrate the benefits (in this case) of a longer flange distance and thus smaller incident angles. But why bother to prove something we already know? Why would we care about how lenses designed for a longer flange distance wouldn’t create color shift issues when we are actually interested in solving color shift issues with existing M lenses?

[AlanG]

Speculation, conjecture, or observation? I am not aware of color shift issues with Kodak’s MF sensors and I don’t see why the design of Kodak’s MF sensors should lead to color shift.

 

 

Kodak’s full-frame DSLRs didn’t even use Kodak CCDs but CMOS sensors designed by Fill Factory in Belgium.

 

 

Hasn’t Jaap already answered the question? The relevant difference between a DSLR and the M9 is the flange distance. Adapting a T/S lens for a Canon EOS to the M9 would demonstrate the benefits (in this case) of a longer flange distance and thus smaller incident angles. But why bother to prove something we already know? Why would we care about how lenses designed for a longer flange distance wouldn’t create color shift issues when we are actually interested in solving color shift issues with existing M lenses?

 

We're all speculating except for the guy who actually put the CV 15 on the 5D. Since I don't have a Nikon mount CV 15 myself, I can't try one. And I'm not going to buy an M9 and adapter to try my 17 TSE on it. But I encourage others to try it and see if we can learn something.

 

You seem to argue points that were never suggested. I didn't say that the Kodak full frame camera had a Kodak CCD chip. Only that it had problems with color shift. The point was that differences between sensors do exist regarding color shifts. And if back then, Kodak couldn't make a better fullframe sensor or spec one from another source, and Nikon and Sony didn't have a full frame sensor until years later, that indicates to me that Canon had superior technology back when the 1Ds came out and for at least several years thereafter. Apparently Sony figured out how to do it, but I am not absolutely sure if Kodak has as I am perplexed that no other 35mm camera companies chose Kodak chips for full frame models.

 

To my way of thinking, Jaapv has not answered the question because he did not say what focal length lens was used, nor if it was tilted as well as shifted. It is valid irrespective of the flange to film distance difference between the two cameras because if the 15mm CV lens works on the 5D but not so well on the M9 that would indicate that the Canon sensor has less or no problem with color shift. (The lens being the same rangefinder type design for both cameras.) How do you explain this other than that the Canon sensor has less propensity to color shift with an extreme w/a non-retrofocus lens than does the M9? I guess another good test would be an old Nikon 21mm F4 used on a Nikon or Canon. As that lens also goes way back into the camera.

 

Also if the Canon 17mm TSE performs poorly on the M9, that would be telling. If it performs well on the M9, then perhaps the M9 sensor is as good as can be made today regarding color shift. I certainly haven't seen anyone who has provided controlled test samples to indicate this. So I am speculating that perhaps the M9 sensor is more sensitive to color shift than the Canon, Nikon and Sony sensors are. Testing would be the only way to establish if this is so or not.

 

You say you are not aware of color shifts on MF sensors. That surprises me as it is a well known situation. Medium format cameras that employ movements (and some fixed wide angle lenses) often require shooting a color compensating target. A few years ago, the rep from one MF back told me that Dalsa chips have fewer problems with color shift. I also have it on good authority from another source at a different MF back company. (Could this have been why Leaf kept using Dalsa chips even after Kodak bought the company? Yes I'm speculating.)

 

Here's the info. from Phase One on shooting lens cast corrections (LCC.) So the chips certainly are prone to color shift.

 

http://xchange.phaseone.com/home/Content/Support/article.aspx?articleid=2118&languageid=1

http://xchange.phaseone.com/KBFiles//2118/1/LCCWin.pdf

[meino]

Alan, just estimating how much you are tilting an SLR lens is not persuasive. The angle of incidence AT THE SENSOR is what counts.

 

The long register distance reduces the basic angle of incidence for a Canon SLR 17mm substantially compared to a Leica RF 18mm. I.E. the Canon lens, sitting out on the front of a mirror box may have an angle of incidence (not angle of view - different things) for the corners of the frame of 20° (without tilt or shift), while the Leica 18mm is probably more like 35°. Lots of leeway for tilt before even getting close to the challenges the Leica sensor has to handle.

 

If someone has a real knowledge of the location of the exit pupil, nodal point, or whatever it is that is the critical spot from which to measure AoI for both lenses, they can no doubt calculate the precise angles involved and do a more exact diagram.

to get an estimate of the exit pupil/ nodal point, use a slide gauge and measure the size of the diafragma at the backside of the objective, and then multiply it with the diafragma setting. This is not an exact measurement, but it can give some aproximate value.

For instance doing this on some of my lenses, I got the following aproximate values

Zeiss 18mm Distagon f4, aprox 25mm

Elmarit ASPH 24mm f2.8, aprox 30mm

Summicron (IV) 35mm f2, aprox 38mm

Summicron 50mm f2, aprox 42mm

Summicron ASPH 90mm f2, aprox 60mm

In my case, the only lens that exhibits the red shift, is the Zeiss 18mm. so I have the feeling that redshift starts when the exit pupil comes below the 29-30mm

[mjh]

Advertisement (gone after registration)

I didn't say that the Kodak full frame camera had a Kodak CCD chip. Only that it had problems with color shift.

…the implication being that it was somehow a problem associated with Kodak. But the color shift issue with Kodak’s DSLRs prove quite the opposite, namely that similar problems can be observed with cameras using different sensors from different vendors (a Kodak full-frame transfer CCD in the case of the M9 vs. a Fill Factory CMOS sensor in the case of Kodak’s full-frame DSLRs).

 

And if back then, Kodak couldn't make a better fullframe sensor or spec one from another source, and Nikon and Sony didn't have a full frame sensor until years later, that indicates to me that Canon had superior technology back when the 1Ds came out and for at least several years thereafter. Apparently Sony figured out how to do it, but I am not absolutely sure if Kodak has as I am perplexed that no other 35mm camera companies chose Kodak chips for full frame models.

Kodak has offered both bigger and smaller than 36 x 24 mm sensors for many years and there is nothing special about the 35 mm frame size. There have been 36 x 24 mm CCDs in Kodak’s portfolio for quite some time, but these are interline transfer CCDs for industrial applications; you are not likely to see one of those sensors (the KAI-16000 for example) in a DSLR.

 

To my way of thinking, Jaapv has not answered the question because he did not say what focal length lens was used, nor if it was tilted as well as shifted. It is valid irrespective of the flange to film distance difference between the two cameras because if the 15mm CV lens works on the 5D but not so well on the M9 that would indicate that the Canon sensor has less or no problem with color shift.

That part is actually conjecture at best.

 

How do you explain this other than that the Canon sensor has less propensity to color shift with an extreme w/a non-retrofocus lens than does the M9?

Apart from the fact there isn’t anything to explain (we are told that there were serious vignetting issues with this lens, just as we would have expected) the retrofocus/non-retrofocus dichotomy has no bearing on this issue. It is the position of the exit pupil and especially its distance from the image plane that counts. Since the flange distance of an SLR sets a lower bound on this distance, lenses for DSLRs will generally be less problematic.

 

Here's the info. from Phase One on shooting lens cast corrections (LCC.) So the chips certainly are prone to color shift.

Only this is an entirely different issue.

 

For all we know, the color shifts observed with the M9 are due to the filter characteristics of the IR cut filters being dependent on the incident angle. The sensor proper, i.e. the chip, is not involved. There are different kinds of filters in use – absorption filters, dichroic filters, and hybrid filters being a sandwich comprising both kinds of filters. Leica uses absorption filters whereas Canon prefers hybrid filters. But all of these filter designs can in principle create color shift issues. A dichroic filter has a transmission curve with a cut-off wavelength that depends on the incident angle: the greater the angle (measured against the optical axis) the more the filter will cut into the red part of the visible spectrum. The transmission curve of an absorption filter doesn’t change with the incident angle, but the filter˚s effectiveness varies with the angle. An absorption filter effectively suppressing IR will also suppress some of the red light, and it will supress even more red if the angle is greater and the distance the light travels through the filter increases. So both types of filters and of course the hybrid designs too can cause vignetting in the red channel and thus cyan corners and edges, and overcorrection of this effect will create red corners and edges. Short of a perfect compensation of the color shift in software the only way to avoid it is to limit the range of incident angles (which is not an option if existing lenses designed for rangefinder cameras are to be used).

[lars_bergquist]

Michael -- how the devil can a plain absorption filter cause a magenta shift at one short edge of the sensor, and a cyan shift at the other? Should not the effect be symmetrical?

 

This dumb old man cannot really figure this out (I do still suspect tha it is all caused by the microlenses) but I do nevertheless think that there are some explanations that should be excluded!

 

The confused old man from the Age of Plain Quantum Physics

[Nikkor AIS]

 

 

Lecia M9 with 21 F4 Super Angulon.

 

 

Most of this discussion is over my head:p. However, I did notice some red edge on a M9 I tested on my 21 Super Angulon. And I didn't like it. And to me it's unacceptable. Clearly the sensor and the lens in front of it are not working for all of the lecia glass that is in perfect for the film M's. My theory is that lenses like the 21 F 4 that extend far into the body and end up close to the Sensor are a no go for the Leica M9.

 

 

Leica M7 and 21 F4 Super Angulon on Ektar 100 film

[Jerry_R]

lenses like the 21 F 4 that extend far into the body and end up close to the Sensor are a no go for the Leica M9.

Would be good, if Leica would finally and officially present list of such lenses, maybe even with two categories, like small shift, and very visible shift.

 

Why users have to loose their time discussing it, makeing several tests, buying & selling lenses - instead of paid engineers in Solms?

[Nikkor AIS]

Or another solution is to forget the M9 and start/continue are use/collection of Leica M film camera's. Or just buy the top of line new lecia glass like the Leica 24 1.4, Leica 20 1.4 for the wide work with the M9

I think people forget that the M9 is a first generation full frame digital leica. Im sure the problems with the red edge will get sorted out in future models of the full frame lecia (M10, M11) but for now it's clear that the sensor/lens in front of the sensor are less than perfect solution. While soft wear fixes might help it's less than a perfect solution.

It's funny that the older Nikkor glass was given new life with the release of Nikon D3. And the camera even corrects for a certain amount of CA and vignetting. So perhaps there is still hope to get the most out of older wide angles like my Super Angulon 21 f4 which is really great on the film M and dismal on the leica M9.

 

 

Taken with Nikkor 45 PC-E 2.8 ED on D3 with SB-900 bounced off roof off camera with TTL cord.

 

Gregory

[adan]

The Super-Angulons have NEVER been supported for use on digital Leica Ms - they are not on the list of lenses that can be retrofitted with coding, Leica has never assigned codes for them, and does not give corrections for them in the M9 menus.

 

(For that matter, they have not been FULLY supported by film Leicas going back as far as the M5 in 1972 - no metering available. They are a 40-year-old design that Leica moved on from 30 years ago with the meter-capable Elmarit)

 

If you want to put them on the camera and see what you get - that's fine. But Leica never said they'd work acceptably.

 

That is different from lenses that ARE supported by lens codes (and do not sit as close to the image plane as the SA) - but still show red edge problems.

 

BTW - try using an old Nikkor 21 f/4 on a D3 - the kind that requires mirror lockup and sits close to the image plane, just like your Leica SA. Then complain to Nikon about the output of that lens on a digital sensor - and watch them try to keep straight faces and not ROTFLTAO.

 

Old 2.1cm f/4-16 Nikkor-O lens

 

The basic fact is that any true, symmetrical WA lens - Nikkor, Super-Angulon, the "real" Zeiss Biogons (21 and 38), or the Zeiss Hologon 15/16 mms - is the exact kind of optical design LEAST likely to be digitally friendly.

 

As Hasselblad themselves will tell you regarding their CFV digital backs: "Not recommended for critical work together with SWC models and ArcBody due to optical incompatibility."

 

CFV-39

[Jerry_R]

it's clear that the sensor/lens in front of the sensor are less than perfect solution

Pity, no info in marketing PDFs...

 

Not to you Gregory, but general comment:

 

Sometimes reading the forum, I have feeling, that BUYERS should know, be aware of so many things...

 

Do you forget, that interest in M9 is huge among people outside the system? Who have no idea about history of "M"?

Don't you think - that more official statement from Leica would minimise confusion to such people?

 

What Leica is showing - is one of two:

- it does not control the situation

- it does control, but is aware that saying truth would stop people from buying, would cause negative press

 

I can't believe, that company with such long experience is hiding its face by saying nothing...

And indeed, I never saw so many forgiving customers... Even worse, they try make excuses in the name of Leica...

 

Doesn't Toyota come to your minds?

[UliWer]

.... But Leica never said they'd work acceptably. ....

 

 

The M9-manual says, the Super-Angulons can be used - though without light-metering (and without coding).

[jaapv]

But if they cannot be used coded it is not possible to implement correction software, so one cannot expect them to be corrected.

[pgk]

I can't believe, that company with such long experience is hiding its face by saying nothing...

Odd that neither Nikon nor Canon say anything about their full frame dSLRs being huge, bulky and heavy;). You are blowing this out of all proportion. There is NO perfect camera built by any manufacturer and no manufacturer is going to discuss the relative merits and compromises of their particular approach to problems. I do not as yet have an M9 but I've seen nothing in this discussion which means that I would not consider buying an one, which, as finances permit, I will probably do (I'm very satisfied with my M8 still though!).

[Jerry_R]

I've seen nothing in this discussion which means that I would not consider buying an one, which, as finances permit, I will probably do (I'm very satisfied with my M8 still though!).

Yu do not belong then to one of potential buyers group, who start looking at this forum AFTER buying camera and starting to see sth strange

(sometimes, not always of course)

[UliWer]

But if they cannot be used coded it is not possible to implement correction software, so one cannot expect them to be corrected.

 

Since we know that Leica is usually very careful declaring usable lens-desgins which were not made for the digital M (e.g. collapsable lenses or the 135mm Apo-Telyt wide opened) it wouldn't be surprising if there was some caveat about the results of a Super-Angulon with a M9.

 

When I look at the examples in # 149 and when I believe they are real (which I still can't believe) I can only say: times when every lens for the screw-mount or M-System was usable on ever M Camera are over. The Super-Angulon does not seem to be usable, even if the manual says it is.

[Advertisement]

Hallo,

Look here: Leica M

[adan]

1. The 21 S/A samples in NikkorAIS's post are accurate. Local store has one, and that's what I got, too.

 

2. Except for B&W, of course. According to the folks crying for a monochrome Digital M, serious Leica photographers shoot only B&W. I don't believe it, myself, but there it is.

 

3. Every day has its sunset.

[noah_addis]

Odd that neither Nikon nor Canon say anything about their full frame dSLRs being huge, bulky and heavy;). You are blowing this out of all proportion. There is NO perfect camera built by any manufacturer and no manufacturer is going to discuss the relative merits and compromises of their particular approach to problems. I do not as yet have an M9 but I've seen nothing in this discussion which means that I would not consider buying an one, which, as finances permit, I will probably do (I'm very satisfied with my M8 still though!).

 

Well, you can easily see the size of a Canon or Nikon SLR by looking at it. Same with the weight, even if you can't see one in person you can check the published specs.

 

Nowhere in the published specs for the M9 did it say that my currently manufactured, coded 21 Elmarit Asph would not perform well on the M9. I can understand Leica not fully supporting old, out of production lenses. And I can certainly understand them not fully supporting third-party lenses. But Leica should have been honest about the camera's performance with lenses in the current lineup.

 

You're right, no camera is perfect. But for the prices of current Leica gear, I don't think it's wrong to expect technical perfection when the camera is used within it's design parameters. The M9 isn't going to shoot 10FPS or ISO 64000. We all knew that by looking at the specs. But we did not know that a technical flaw or incompatibility would give poor performance with some current Leica lenses.

 

Leica needs to speak up, and fast. I need an honest answer about what's causing the problem, if it's fixable and what they're going to do about it if it's not.

 

I'm fed up. Can't even get anyone at Leica to return my calls. Again, that's the kind of service I'd expect from the larger manufacturers. I foolishly expected more from Leica.