You're mistaken on most of it. I assumed one would always use an IR filter on the M8 unless shooting B&W. You do still get cyan drift on the M8 with no filter if you use a symmetrical wide angle, but to a lesser degree than with the filter.
On the M9 coding is very much for colour correction, the illumination vignetting on the sensor with a totally uncoded lens is very similar to film, one of the results of the extreme microlens shift.
Why would they exhibit colour-vignetting on film
Here is a diagram that might clear it up (not to scale ):
Thicker the glass, the shorter the wavelengths it will absorb (think water getting greener as it gets shallower and eventually clear, but deeper blue as it gets deeper).
So the centre doesn't cause a colour shift as it is calculated to just block the IR wavelengths.
Because of the effectively thicker glass light has to pass through toward the edges if the lens has a high angle of incidence, red light is absorbed toward the edges of the image circle we get cyan drift.
So to summarise, the red edge(s) are caused by a SOFTWARE overcorrection for cyan drift that and their SOFTWARE profile isn't perfectly symmetrical.
As i understand it, the major difference with M8 and M9 is which side of the offset microlens layer the IRcut is on. This means that on M9 the IRcut only sees light that has already been corrected for its angular path, and is therefore optimally controlled.
I would think cyan drift issues will be followed up on in due course, as they were with M8. Preliminary views of M9 imagery would suggest that the IR problem has been resolved, thats a good start.
The IR absorbing coverglass was thickened slightly (0.8mm vs 0.5mm) and they use a slightly different material. The microlenses still sit directly above the photodiode on the sensor.
The microlens shift was slightly increased and the red dye used in the CFA was changed, otherwise it is the same sensor tech as the M8.
IRCut filters are thin films, so much thinner than the glass they're put on that the strongest filter won't make the filter any thicker. If the microlens layer is thicker, then it is to do with offsets optimised for FF, not because of a heavier IRCut
thrice wrote:
You're mistaken on most of it. I assumed one would always use an IR filter on the M8 unless shooting B&W. You do still get cyan drift on the M8 with no filter if you use a symmetrical wide angle, but to a lesser degree than with the filter.
On the M9 coding is very much for colour correction, the illumination vignetting on the sensor with a totally uncoded lens is very similar to film, one of the results of the extreme microlens shift.
Why would they exhibit colour-vignetting on film
Here is a diagram that might clear it up (not to scale ):
Thicker the glass, the shorter the wavelengths it will absorb (think water getting greener as it gets shallower and eventually clear, but deeper blue as it gets deeper).
So the centre doesn't cause a colour shift as it is calculated to just block the IR wavelengths.
Because of the effectively thicker glass light has to pass through toward the edges if the lens has a high angle of incidence, red light is absorbed toward the edges of the image circle we get cyan drift.
So to summarise, the red edge(s) are caused by a SOFTWARE overcorrection for cyan drift that and their SOFTWARE profile isn't perfectly symmetrical....Show more →
Oh, I've not seen any examples of cyan corners on the M8 with no IR filter in front of the lens.
As far as the M9, again, I've not seen any examples of color vignetting without coding. The red left side I've only seen on M9 photos WITH correction. Maybe it occurs without correction, but again I've not seen it. And from what I've read, the problem with the software correction is that it IS symmetrical, yet the vignetting is not, either spatial or with respect to wavelength.
Regardless, I posted an example of color drift in the corners in a film shot in this very thread. I don't know why the color vignetting happened, but the corners of the shot look bluer than the rest of the shot to me. It was the one taken with the CV 15/4.5.
And honestly, that is not what causes the bulk of the cyan corners on the M8 with an IR filter in front of the lens. Those IR/UV filters are not simple filters that rely on absorption; they are interference filters. The angle of incidence affects the wavelength that the filters destructively interfere with and don't pass. Try one of those on a film camera at a wide angle focal length like 28mm and see what you get. The hot mirror in front of the sensor has a small effect on cyan corners, but the bulk of it comes from the interference filter in front of the lens.
Rob Riley wrote:
IRCut filters are thin films, so much thinner than the glass they're put on that the strongest filter won't make the filter any thicker. If the microlens layer is thicker, then it is to do with offsets optimised for FF, not because of a heavier IRCut
M9 doesn't use a hot mirror, so the first point doesn't apply. It uses an absorption filter.
The microlens layer isn't thicker, it is the same. The IR absorption filter that acts as coverglass is thicker. I have that straight from Stephan Daniels' mouth (product manager for digital M at Leica).
That's a great post from Joe, but the M8/M9 don't use an IR cut filter.
kidtexas, I think the vignette on that 15/4.5CV shot is the result of the film/developer/scanning/colour-treatment. Otherwise this sort of behaviour would be documented and have a scientific foundation. Unless you can submit such an explanation forgive me if I don't accept one example as gospel.
I don't want to get in trouble, but as I assume you're not a Reid Reviews subscriber i'm going to share part of his 15mm CV review with you, here is a vignetting analysis without a filter on the lens, taken on the M8: http://i9.photobucket.com/albums/a96/Nephilim666/Cyan_Drift2.jpg
I'd say that reliably puts the argument to bed.
kidtexas wrote:
As far as the M9, again, I've not seen any examples of color vignetting without coding. The red left side I've only seen on M9 photos WITH correction. Maybe it occurs without correction, but again I've not seen it. And from what I've read, the problem with the software correction is that it IS symmetrical, yet the vignetting is not, either spatial or with respect to wavelength.
I never said (or implied) red corners occur without correction.
If the sensor isn't perfectly aligned to the lens flange then yes, symmetrical correction profiles would cause this problem, that remains to be proven in testing or measurement. I'll ask one question though, why doesn't it occur with the WATE if it's a sensor alignment issue?
Im not surprised Daniels would say that, Leica actually published such a statement sometime around the M8 crisis, then one day it was gone. I had an enormous argument with Joe W based on that statement from Leica, and I lost b/se that analogy is erroneous. I couldnt find 'that' argument (they probably blew it away) but I did fall over the following
"The comments on the M8's IR performance at Luminous Landscape
(http://www.luminous-landscape.com/reviews/cameras/m8-infrared.shtml) say that Leica was forced to use a very thin (and presumably less efficient) IR blocking filter because they were worried about a thicker filter introducing color fringing when working with wide angle lenses. The result is the IR leakage that people are complaining about."
and then from Joseph Wisniewski
"That is pure BS. The filter may have been weaker (as in some of my other posts) to get around an angle affect peculiar to rangefinders, but it isn't any thinner. Modern IR filters are thin films, so much thinner than the glass they're put on that the strongest filter won't make the filter any thicker. "
which is i believe, the same thing
I fell over these posts on the way to finding the above
Yet you still don't get it. IT DOESN'T HAVE AN IR CUT FILTER!
Do you know the difference between IR absorbing glass and an IR filter? If not, please stop trying to argue with someone who understands the technologies at work in the camera being discussed.
The reason for the M8 having the IR issues is because they chose to have a THINNER IR-ABSORBING coverglass in order to maximise sharpness with extremely oblique incident rays that occur in rangefinders, I can explain that in optical theory to you as well if need be. Do I need to tell you how IR absorbing glass works as well? Joe is merely incorrect in this case, obviously he hasn't done enough research into the camera itself. I trust Leica engineers for an explanation more than I trust some guy on a forum.
How they overcame it in the M9 is through using a better glass type for the absorbing filter allowing them to increase the thickness by 0.3mm and giving an effective enough IR filtration without lowering sharpness. The cyan drift is more severe because of the more effective filter, why we haven't seen any examples (this point is for kidtexas) is because people can select profiles for their uncoded lenses in the menu. I'll happily show you some examples in about a week.
It would depend on which lens is being used insofar as it would have to produce oblique incident rays landing on the sensor. As far as I'm aware there are no telecentric wide-angles for 35mm rangefinders so my statement doesn't need any further clarification, but feel free to correct me on that.
However, both will cause cyan drift... I think it would be too laborious to list every lens in order of severity. I don't think I understand exactly what you want me to do to make the argument more sensical.
thrice wrote:
However, both will cause cyan drift... I think it would be too laborious to list every lens in order of severity. I don't think I understand exactly what you want me to do to make the argument more sensical.
I don't want you to do anything but accept that you can't say that every lens of this or that FL will give significant "cyan drift". In that case it would be the same on any camera, even SLR:s, since there are no 100% telecentric lenses for those either.
You can put a 24 mm EOS lens on the M9, with an adaptor. Will it give less "cyan drift" than the M-lenses? Yes, most likely. Will it act different in this respect than when it is sitting on an EOS camera? No, of course not. It's the same lens with the same "telecentricity".
thrice wrote:
kidtexas, I think the vignette on that 15/4.5CV shot is the result of the film/developer/scanning/colour-treatment. Otherwise this sort of behaviour would be documented and have a scientific foundation. Unless you can submit such an explanation forgive me if I don't accept one example as gospel.
I don't want to get in trouble, but as I assume you're not a Reid Reviews subscriber i'm going to share part of his 15mm CV review with you, here is a vignetting analysis without a filter on the lens, taken on the M8: http://i9.photobucket.com/albums/a96/Nephilim666/Cyan_Drift2.jpg
I'd say that reliably puts the argument to bed.
I never said (or implied) red corners occur without correction.
If the sensor isn't perfectly aligned to the lens flange then yes, symmetrical correction profiles would cause this problem, that remains to be proven in testing or measurement. I'll ask one question though, why doesn't it occur with the WATE if it's a sensor alignment issue?...Show more →
The corners I showed are not due to scanning or color treatment - it's there on the slide. It could be Kodachrome's shadow rendering - it might shift to the blue. I don't know. However, I've seen it on many of my CV 15/4.5 shots on shot on Portra, so it's not limited to Kodakchrome. I didn't claim an explanation for it. I have it in several shots. It's either the due to the lens design and how it transmits light, or the way vignetting affects color on these films.
No, your picture does not put that argument to bed. I agreed with you that the IR filter over the sensor can cause cyan corners, but again, the bulk of the problem is due to the IR filter over the lens. I had a discussion with Sean Reid about the IR cut filters; he didn't understand how they worked at the time either. I don't know if he ever figured it out since I didn't follow up on the discussion. The CV 15/4.5 might get cyan corners without a lens mounted IR cut for the reason you state, due to its more symmetrical design, but almost ALL lenses greater than 35mm or so get really cyan corners with the IR filters.
I've been following the M8 since it came out, and it's well documented that without IR filters, your shots have IR contamination. The solution to that, adding IR filters on the lenses, caused another problem, cyan corners. It's in many of the reviews. Some lenses might additional have slight cyan corners due to the reasons you state.
I played around with an M9 yesterday. One of the things I specifically did was disable lens detection and look for color issues. I saw vignetting, but no cyan corners.
Lastly, look around for why Leica doesn't have corrections for the UV/IR filters on the M9. It's because the shift in the filter pass frequency, starting with 75mm or 50mm lenses, gets so extreme with wide angles that possibility of proper correction is low.
I said nothing about sensor alignment issues either. I don't know what causes the asymmetrical vignetting on the M9 and why symmetrical profiles don't correct for it. The best testing I've seen on the matter is by Sandy of Chromasoft, which I linked to earlier. It's also not red corners. It's a red left side.
I don't really have the time or desire to continue this. IR cut filters in front of the lens, as used on the M8, are interference filters. I use interference filters everyday at work. They block at the frequency they are design for at normal incidence. They cause the bulk of the cyan corners on the M8. The M9 should be largely free from this problem.
Yes abam I answered that question in my first post, heaven forbid a thread that was destined to go nowhere get diverted toward a scientific discussion of the camera... on a gear forum. You know where they irrationally whine or gush over gear? forum 2.
kidtexas, I don't disagree with anything you just wrote. Sorry if you feel I implied that you said the camera might have sensor alignment issues, it was just one hypothesis. Some films can produce colour shifts in shadows, I know Velvia 50 does, I don't know if these shifts can occur in vignetting but I accept that it may be due to the lens as well. I have a 25 biogon that is rather symmetrical, I'll hopefully be able to test it's effect on the M9 soon.
):
Two 24 mm lenses of different brand will of course not have the exact same exit pupil distance to the image plane.