A Sane Attitude to Rededge

[adan]

Advertisement (gone after registration)

I'm glad to hear that the coded or manually selected corrections for the 21 pre-ASPH at least work for someone's lenses - too bad they don't work perfectly with my 21 pre-ASPH.

 

=8^o

 

BTW - I will just mention that my M9 does not suffer from any of the other bugs or problems people have reported - possibly because of my work style (no auto-review, very little reviewing at all, very little dependence on "auto-off"). It's a glorious tool - except for the darn red edge with my most-used lens.

[lct]

I'm glad to hear that the coded or manually selected corrections for the 21 pre-ASPH at least work for someone's lenses - too bad they don't work perfectly with my 21 pre-ASPH...

Did you try the 21/2.8 asph Andy? Good lens no?

[ho_co]

Suggestion as to possible origin of red-edge (my take) from another thread:

 

 

... A video of the assembly of an M9 shows that sensor chips are trued by putting shims around their perimeter. These shims may deform the sensor and predispose its IR glass to crack. If this is the cause, it suggests that even if the IR glass does not crack, the sensor may still be slightly deformed on countless M9's.

[vanhulsenbeek]

Suggestion as to possible origin of red-edge (my take) from another thread:

 

 

Quote:

Originally Posted by Hey You

... A video of the assembly of an M9 shows that sensor chips are trued by putting shims around their perimeter. These shims may deform the sensor and predispose its IR glass to crack. If this is the cause, it suggests that even if the IR glass does not crack, the sensor may still be slightly deformed on countless M9's.

 

Hm, I would doubt that. Here is a page from Mark Nortons' immortal series Anatomy of an M8.

Presuming a similar setup is used, it seems the shims are some distance from the sensor itself, even in the M9, and I doubt the sensor bottom plus the mounting board are so weak that they would warp:

 

http://www.l-camera-forum.com/leica-forum/leica-m8-forum/21331-anatomy-leica-m8.html#post224641

[mjh]

Using shims for achieving proper alignment is standard practice; I’m not convinced this is responsible for the cracked cover glass issue. Now it cannot strictly be ruled out, especially since the glass is thicker and thus less pliable than the one on the M8’s sensor, but still – if this is done right, there shouldn’t be any undue stress on the sensor and its cover glass.

[charlesphoto99]

I've always coded my Zeiss 18mm as a 21 pre-asph. Worked like a charm on the M8. I found the Wate setting to introduce red edges, at least withe the B&w uv/ir I was using. Honestly haven't even mounted it on the M9. Just never thougt of M's as superwide cameras, 24 about as wide as I like. If I need superwide out comes the Nikon 14-24 which is a superb lens.

[jaapv]

Advertisement (gone after registration)

Using shims for achieving proper alignment is standard practice; I’m not convinced this is responsible for the cracked cover glass issue. Now it cannot strictly be ruled out, especially since the glass is thicker and thus less pliable than the one on the M8’s sensor, but still – if this is done right, there shouldn’t be any undue stress on the sensor and its cover glass.

This is my theory, Michael, based on the reasons why dental composites sometimes crack ( another thin flexible layer application). Tell me whether it's nonsense or not...

 

I don't think so.My personal theory is that the problem is that the thickness of the cover glass, which is thinner that a DSLR cover glass and thicker than the M8 cover glass ( and double the surface area of the M8) puts it just into a flexibility range that can cause breakage during thermal cycling, which occurs in a CCD sensor during exposure. The flexibility of the intermediary kit layer is supposed to compensate the different rates of expansion. If there is just the slightest discrepancy in either the thickness or the viscosity of the kit, this can cause breaking tension. I very seriously doubt that this is the cause of the red-edge problem. Another FF sensor that had this " Italian Flag" problem, even a lot worse than the M9, the Kodak 14 DSC, never suffered broken IR filters.

[mjh]

This is my theory, Michael, based on the reasons why dental composites sometimes crack ( another thin flexible layer application). Tell me whether it's nonsense or not...

One would have to do the maths (if one only knew the values of the relevant parameters) but it sounds plausible. The Kodak DCS Pro 14n featured a CMOS sensor (so there was less heat) and a thick IR absorption filter (thus the “italian flag” issues that seem to be the bane of thick absorption filters).

[ho_co]

Clearly, I may be quite wrong. Thanks, Michael, for reminding me that shims are old hat, but we've just started having the broken sensor glass problem.

 

Here's part of a suggestion I posted in the broken sensor glass thread:

Here's a test for the theory:

 

Do a comparison of the two groups:

1) people who have red edge syndrome; and

2) people who have broken cover glass.

 

It would fit the theory that both are caused by the same thing if people who have the red edge problem later end up with a cracked cover glass. (The louder you complain of red edges, the quicker your coverglass breaks. )

...

 

NOTE: I'm not making an accusation, but suggesting a test for one set of assumptions. We're all curious about why the red edge syndrome is so variable, and we're all curious about why a small number of sensors have cracked. If we build a single database of both phenomena, we can get statistical evidence for or against the idea that the two might be related.

[meino]

A question, is the cover glass always broken in the same location?

I saw some examples where it was broken in the lower left corner. That is also the location where the third mounting lug is located and where the red edge phenomen is at its worst.

[SJP]

Executive summary: red-edge problem is due to - (1) high angle of incidence, (2) full frame sensor size and (3) (very slightly asymmetric) alignment of microlenses w.r.t. the pixel bins

 

No.3 is the real challenge as this will vary from sensor to sensor so you cannot make a one size fits all solution. This would explain why some camera's are performing better than others.

 

The best way of dealing with this is to have a built in "Cornerfix like" algorithm i.e. measure and compensate. Alternatively Leica might be able to offer an automatic fix to bulk process the DNG's (or jpegs) i.e. auto cornerfix on importing of files.

 

Finally I expect that the red-edge correction even depends on the point focus (infinite versus close), focussing shifts the lens forward and back and therefore changes the angle of incidence. Anyone willing and able to test that? Preferably with a non-retrofocus lens (SA or similar).

[AlanG]

Executive summary: red-edge problem is due to - (1) high angle of incidence, (2) full frame sensor size and (3) (very slightly asymmetric) alignment of microlenses w.r.t. the pixel bins...

 

 

It seems to me that all of these things have to do with the angle that the light hits the sensor. So why is it that tilt shift lenses work fine on Canon and Nikon cameras? The color doesn't change any noticeable amount when I tilt or shift these lenses which makes the angle the light hits the sensor change quite a lot. (Despite the sensor being a bit farther back on the DSLRs.... a 12mm shift combined with a full swing is a pretty significant movement for a 17mm lens.) And my Nikon 35mm and 28MM PC lenses along with the 24mm Canon TSE had terrible color shifts when used on the old full frame Kodak DCS SLRc camera.

 

I think Kodak and Leica just haven't figured it out yet.

[vanhulsenbeek]

Or those forum members that do not see, it or are not bothered by it, have not figured it out?

[viramati]

Very surprised to see this shot with 28 cron ASPH at probably f5.6 after upgrade. I tis there on the right which would have been the top of the frame from the way I hold the camera

 

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!

 

Have also done the typical white wall shot and there is definite signs of pink at left and along top of frame!! Had never noticed this before so is it a step backwards for this particular lens?

[jaapv]

It seems to me that all of these things have to do with the angle that the light hits the sensor. So why is it that tilt shift lenses work fine on Canon and Nikon cameras? .

Because the register is so long that the angle is not particularly acute, even with a shift lens.

[AlanG]

Because the register is so long that the angle is not particularly acute, even with a shift lens.

 

I do not believe you are correct. I think Canon and then Nikon (Sony) simply make full frame sensors that are not as affected by the angle.

 

I looked at my lenses very closely and the angle change is very substantial when you shift or tilt. Tilt alone applies as much as 8.5 degrees of change and this is irrespective of register distance. Combining maximum tilt and shift and one is increasing the angle by nearly 20 degrees. And as I said previously, if the register distance solves the problem, then why did the color shift so much when I used the lenses on the Kodak full frame cameras when I employed shifts and tilts? Clearly some sensor are much more sensitive to this than others.

 

Also see this:

 

Voigtlander 15mm v Sigma 12-24mm: Intro

[jaapv]

I do not believe you are correct. I think Canon and then Nikon (Sony) simply make full frame sensors that are not as affected by the angle.

 

I looked at my lenses very closely and the angle change is very substantial when you shift or tilt. Tilt alone applies as much as 8.5 degrees of change and this is irrespective of register distance. Combining maximum tilt and shift and one is increasing the angle by nearly 20 degrees. And as I said previously, if the register distance solves the problem, then why did the color shift so much when I used the lenses on the Kodak full frame cameras when I employed shifts and tilts? Clearly some sensor are much more sensitive to this than others.

 

Also see this:

 

Voigtlander 15mm v Sigma 12-24mm: Intro

As it happened I was playing with a Nikon shift lens on the M9 last Friday, and there was not a trace of red edge, even with the lens at its max 11 mm. I regret I did not keep the files, so you will have to take my word for it. I will ask AJ55 to take the combo when I see him which will probably be the 28th and take some new images.

[AlanG]

As it happened I was playing with a Nikon shift lens on the M9 last Friday, and there was not a trace of red edge, even with the lens at its max 11 mm. I regret I did not keep the files, so you will have to take my word for it. I will ask AJ55 to take the combo when I see him which will probably be the 28th and take some new images.

 

That is interesting especially in light of the result of the CV 15 on a 5D. I suggest that next time you try the lens tilted and shifted with a white diffuser over the lens. (Plexiglass, opal glass, etc.)

[scott kirkpatrick]

As it happened I was playing with a Nikon shift lens on the M9 last Friday, and there was not a trace of red edge, even with the lens at its max 11 mm. I regret I did not keep the files, so you will have to take my word for it. I will ask AJ55 to take the combo when I see him which will probably be the 28th and take some new images.

 

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

[Advertisement]

Hallo,

Look here: Leica M

[AlanG]

As it happened I was playing with a Nikon shift lens on the M9 last Friday, and there was not a trace of red edge, even with the lens at its max 11 mm. I regret I did not keep the files, so you will have to take my word for it. I will ask AJ55 to take the combo when I see him which will probably be the 28th and take some new images.

 

By the way, you didn't say which Nikon lens you are referring to. Was it the 24mm tilt shift?