Crack!

[wlaidlaw]

Advertisement (gone after registration)

Jono and others,

 

What you have all been waiting for with bated breath - White Wall, the Sequel with WATE at 16mm/f4. This time with lens detection set to off. The vignetting is as to be expected, worse, as is red edge. However, it is just as asymmetrical, so it is not the camera providing asymmetrical correction. I just don't know the root cause of this. If Leica say to us, with no equivocation, that the sensor is central, then we must believe them. Other causes - pass.

 

Wilson

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]

Jaap--

Actually, in that case, Sony acknowledged the defect and agreed to replace that batch of defective sensors at no charge to the manufacturer.

 

But I agree that we haven't anything to worry about.

 

Consider the cracking M8 bodies and the self-destructing shutters. Stefan Daniel has said they look at those cases on an individual basis.

 

In my experience, Leica is quite fair to its customers. After all, we're all they've got.

I knew somebody with a Panasonic with the same sensor - it took him a dirty fight with lawyer threats to get it replaced - and only by quoting Leica's example. There is a difference between companies.

[jonoslack]

Jono and others,

 

What you have all been waiting for with bated breath - White Wall, the Sequel with WATE at 16mm/f4. This time with lens detection set to off. The vignetting is as to be expected, worse, as is red edge. However, it is just as asymmetrical, so it is not the camera providing asymmetrical correction. I just don't know the root cause of this. If Leica say to us, with no equivocation, that the sensor is central, then we must believe them. Other causes - pass.

 

Wilson

 

Thanks Wilson - should really have done it myself shouldn't I!

Just as clear . . . . . . except that now there is a clear red tint on the left . . . and just as clear blue-green tint on the right hand side. I can't see how a misaligned sensor could possibly be responsible for that (can you?). Added to this, the area which is darker on the left extends much further than could possibly be explained by a mm or 2 shift of the sensor (don't you think?).

 

Clearly something is happening, and the only thing I can think of which could be doing that is the theory about the asymmetrical nature of the bayer filter - and the two greens for one red and blue might easily account for the fact that the vignetting is less on the right.

 

Of course, we don't HAVE to believe Leica about the off centre sensor - but I'm inclined to believe them because I can't really see how the misaligned sensor theory accounts for the observable problem.

 

I believe this is also observable on some MF cameras (which again mitigates against the misaligned sensor theory). It was certainly an issue on the Kodak SLR/n, where the 'Italian Flag' effect was a real issue with some lenses (and there was no misaligned sensor there either). I was working on the beta software for that with Kodak, I don't believe that anyone really understood the issue then, and I've seen no evidence that it's properly understood now either. All the efforts were to mitigate the effect rather than discover the source. (of course, the M9 has a Kodak sensor, but the Kodak SLR/n didn't).

 

Of course - it could be two or three different issues combining to cause the problem.

 

I only chipped in because of the categorical statements by Shootist which didn't seem to be correct. From a personal point of view, I simply don't have an issue with the colour shift on my M9, and if I did, I'd be using Sandy's cornerfix to sort it out.

 

all the best

[Nicoleica]

My 2¢ worth would be that it's a prismatic effect caused by the extreme angles of the light rays hitting the cover glass when very short focal length lenses are used.

[ArtZ]

Jono and others,

 

What you have all been waiting for with bated breath - White Wall, the Sequel with WATE at 16mm/f4. This time with lens detection set to off. The vignetting is as to be expected, worse, as is red edge. However, it is just as asymmetrical, so it is not the camera providing asymmetrical correction. I just don't know the root cause of this. If Leica say to us, with no equivocation, that the sensor is central, then we must believe them. Other causes - pass.

 

Wilson

 

Wilson,

 

Your pictures just prove it's time to clean (or to paint) your walls :-) I see them beige, not white :-P

 

Just kidding, Wilson.

 

I was at Le Mans Classic 2010 and I was using a M9 from a friend (the one who has the sensor cracked). Both of us are convinced about the sensor not centered theory. We took quite a lot of pictures with my WATE and with a Zeiss 21/2.8 [(whistling... and looking with wondering eyes... -beautiful lens...)] and we got pictures like yours. On my M8s, vignetting is always consistent: same amount on the left and on the right.

 

Regards

[mjh]

One of the problems with the off-center-sensor-theory has been mentioned before: To account for the observed asymmetry in vignetting, the sensor would have to so far off the center that there would be no debate about this – it would be fairly obvious when looking into the camera with the shutter open. Only it isn’t.

 

Also, the off-center-theory is trying to explain too much. Either the sensor is mounted off-center because it wouldn’t fit otherwise; then it could explain asymmetrical vignetting (though not the amount of asymmetry we are observing). Or Leica tried to force the sensor into a centered position even though there wasn’t enough space, causing stress that is eventually causing cracks in the sensor’s cover glass. That would explain the cracks, but not the asymmetrical vignetting. You cannot explain both.

 

As to the cracks, a faulty batch of sensors is still the most likely explanation, I think. Explaining asymmetrical colour shift, on the other hand, is far from easy. With incident angles growing towards the sides and corners of the image, one would expect red vignetting (and thus a cyan shift) due to the IR-blocking absorption filter (aka cover glass) – more red gets absorbed if the light takes a longer route through the filter. This would be a completely symmetrical effect, though, as would be any prismatic effects caused by the cover glass. Then there’s the bayer pattern: assuming that light passing through a colour filter wouldn’t hit the sensor pixel directly underneath, but a neighbouring pixel supposed to be sensitive to a different colour, that would indeed cause a colour shift, but uniformly towards some shade of magenta (there would be some magenta-shift toward the sides and two diagonally opposing corners; there would be next to no colour shift in the other two corners). I am not aware of a convincing explanation for the Italian flag effect.

[jonoslack]

Advertisement (gone after registration)

 

I was at Le Mans Classic 2010 and I was using a M9 from a friend (the one who has the sensor cracked). Both of us are convinced about the sensor not centered theory. We took quite a lot of pictures with my WATE and with a Zeiss 21/2.8 [(whistling... and looking with wondering eyes... -beautiful lens...)] and we got pictures like yours. On my M8s, vignetting is always consistent: same amount on the left and on the right.

 

Regards

 

Hi there

If you 'cut out' the m8 size from Wilson's image, I think that you'd find that the vignetting was then equal (which, after all, is the case).

 

However, some have seen the colour shift in other cameras (including the M8) which don't have non-centred sensors.

 

But even if it were the case that the sensor was off centre - that wouldn't explain the green/blue on the right hand of the image. I think this whole theory stems from Mark's idea that it was impossible to fit a full frame sensor in an M8 body.

 

What I'm trying to say is that even if the sensor were to be off centre (which I'm not agreeing to) then it wouldn't adequately explain either the colour shift or the vignetting difference. On the other hand the reality of the colour shift does go some way to explaining the difference in vignetting on Wilson's image.

[jonoslack]

Then there’s the bayer pattern: assuming that light passing through a colour filter wouldn’t hit the sensor pixel directly underneath, but a neighbouring pixel supposed to be sensitive to a different colour, that would indeed cause a colour shift, but uniformly towards some shade of magenta (there would be some magenta-shift toward the sides and two diagonally opposing corners; there would be next to no colour shift in the other two corners). I am not aware of a convincing explanation for the Italian flag effect.

 

HI Michael

is it really the case that this would create magenta in both directions? (I bow to your greater judgement).

 

One thing which really does seem to be the case is that nobody understands it properly. Certainly that was the case with Kodak back in 2006 (or whenever the Italian Flag came up). I've talked with technical people who you would really expect to understand it, and they don't seem to know either. I don't see any motive for anyone to keep it a secret either!, after all, most M9 users have either:

 

1. never heard of it

2. not seen it

3. not considered it a significant issue

4. learned to use cornerfix

5. worked out PP fixes when it rears it's ugly head.

6. sold their M9 and bought something else!

 

I don't think anyone is seriously expecting a fix for it - personally I wonder whether it would be fixed in an M10, even if they were to decide on a CMOS sensor (the Kodak SLR'n used a CMOS sensor).

[mjh]

is it really the case that this would create magenta in both directions?

For any sensor with RGB filters arranged in a Bayer pattern it is true that if in one row of pixels there are red-sensitive pixels to the left and right of each green-sensitive pixel, there will be blue-sensitive pixels to the left and right of any green-sensitive pixel in the next row … and so on. The same holds if you are looking at columns of pixels – one column has red pixels on top of and below each green pixel, the next column has blue pixels adjacent to the green ones, and so on.

 

So assuming that near the edges of the image the light passing some colour filter will not hit the sensor pixel underneath, but an adjacent one, it will be green light hittiting pixels nominally sensitive to either red or blue light, and red or blue light hitting pixels nominally sensitive to green. If our test subject was, say, a graycard, a white wall, or something like that, the only difference made by the misguided light rays would be the result of the differences in sensitivity of the red-, green-, and blue-sensitive pixels. Green filters have a higher transmission than the red and blue ones, and together with the sensor’s non-uniform sensitivity for the different wavelengths it usually turns out that the green-sensitive pixels are the most sensitive while the blue pixels have the lowest sensitivity – but the difference between red and blue is generally lower than that between either of those pixels and the green pixels. This difference must be compensated for in the internal image processing or our images would have a greenish tint; i.e. the brightness of red and blue gets enhanced to match the brightness in the green channel. This also explaines why there is usually more noise in the blue than in the green channel. If the assumption that red- and blue-sensitive pixels receive less light and thus accumulate less electrons in their capacitators turns out to be wrong, the enhancement of red and blue results in an excess of red and blue, or in other words in a magenta tint. This is the reason why the dominant colour of several image artefacts is magenta – IR contamination (affecting all the pixels equally, regardless of their colour filter) being one example and smearing (an opto-electronical effect uniformly affecting all the pixels in a colum) another.

 

In this case we have green blue- or red-sensitive pixels, presumed to be hit by the comparatively dim red or blue light, but getting hit by the brighter green light on the one hand, and pixels expecting bright green light but getting dim red or blue light on the other. So with red- and blue-sentive pixels collecting more light than usual and green-sensitive pixels collecting less, the net result is a colour shift towards red and blue – in other words, magenta. The only difference between the left and right edge of the sensor (assuming landscape orientation) is that one side has a red shift in any odd and a blue shift in any even row of pixels, while on the other side the red shift occors in the even and the blue shift in the odd rows. Since the RGB colour of each image pixel is interpolated from neighbouring pixels, this should make no difference whatsoever.

 

The corners of the sensor are a little different, and in fact in my previous post was in error. If the light hits a sensor pixel diagonally adjacent, there wouldn’t be much of a colour shift: light passing through a green filter would actually hit another nominally green-sensitive pixel (just not the one directly underneath) while red light would hit blue-sensitive pixels and vice-versa. Since the difference in the sensitivities of red and blue pixels is comparatively small, the resulting colour shift would be equally small.

[jonoslack]

For any sensor with RGB filters arranged in a Bayer pattern it is true that if in one row of pixels there are red-sensitive pixels to the left and right of each green-sensitive pixel, there will be blue-sensitive pixels to the left and right of any green-sensitive pixel in the next row … and so on. The same holds if you are looking at columns of pixels – one column has red pixels on top of and below each green pixel, the next column has blue pixels adjacent to the green ones, and so on.

 

So assuming that near the edges of the image the light passing some colour filter will not hit the sensor pixel underneath, but an adjacent one, it will be green light hittiting pixels nominally sensitive to either red or blue light, and red or blue light hitting pixels nominally sensitive to green. If our test subject was, say, a graycard, a white wall, or something like that, the only difference made by the misguided light rays would be the result of the differences in sensitivity of the red-, green-, and blue-sensitive pixels. Green filters have a higher transmission than the red and blue ones, and together with the sensor’s non-uniform sensitivity for the different wavelengths it usually turns out that the green-sensitive pixels are the most sensitive while the blue pixels have the lowest sensitivity – but the difference between red and blue is generally lower than that between either of those pixels and the green pixels. This difference must be compensated for in the internal image processing or our images would have a greenish tint; i.e. the brightness of red and blue gets enhanced to match the brightness in the green channel. This also explaines why there is usually more noise in the blue than in the green channel. If the assumption that red- and blue-sensitive pixels receive less light and thus accumulate less electrons in their capacitators turns out to be wrong, the enhancement of red and blue results in an excess of red and blue, or in other words in a magenta tint. This is the reason why the dominant colour of several image artefacts is magenta – IR contamination (affecting all the pixels equally, regardless of their colour filter) being one example and smearing (an opto-electronical effect uniformly affecting all the pixels in a colum) another.

 

In this case we have green blue- or red-sensitive pixels, presumed to be hit by the comparatively dim red or blue light, but getting hit by the brighter green light on the one hand, and pixels expecting bright green light but getting dim red or blue light on the other. So with red- and blue-sentive pixels collecting more light than usual and green-sensitive pixels collecting less, the net result is a colour shift towards red and blue – in other words, magenta. The only difference between the left and right edge of the sensor (assuming landscape orientation) is that one side has a red shift in any odd and a blue shift in any even row of pixels, while on the other side the red shift occors in the even and the blue shift in the odd rows. Since the RGB colour of each image pixel is interpolated from neighbouring pixels, this should make no difference whatsoever.

 

The corners of the sensor are a little different, and in fact in my previous post was in error. If the light hits a sensor pixel diagonally adjacent, there wouldn’t be much of a colour shift: light passing through a green filter would actually hit another nominally green-sensitive pixel (just not the one directly underneath) while red light would hit blue-sensitive pixels and vice-versa. Since the difference in the sensitivities of red and blue pixels is comparatively small, the resulting colour shift would be equally small.

Excellent Michael

Thank you for that.

 

Do you have any interesting theories about what may be happening?

[ArtZ]

Hi there

If you 'cut out' the m8 size from Wilson's image, I think that you'd find that the vignetting was then equal (which, after all, is the case).

 

However, some have seen the colour shift in other cameras (including the M8) which don't have non-centred sensors.

 

But even if it were the case that the sensor was off centre - that wouldn't explain the green/blue on the right hand of the image. I think this whole theory stems from Mark's idea that it was impossible to fit a full frame sensor in an M8 body.

 

What I'm trying to say is that even if the sensor were to be off centre (which I'm not agreeing to) then it wouldn't adequately explain either the colour shift or the vignetting difference. On the other hand the reality of the colour shift does go some way to explaining the difference in vignetting on Wilson's image.

 

Hi Jono,

 

You're probably right when you say "this whole theory stems from Mark's idea that it was impossible to fit a full frame sensor in an M8 body".

 

I haven't bought a M9 mainly because three reasons (in the following order):

 

1.- Too many problems (and too many trips Paris-Solms) until I got my M8s working properly. I wanted to know the problems first.

 

2.- Wanted a real chrome top (not grey paint). I decided to wait because I thought Leica would offer a chrome top.

 

3.- ´Red on the left' and 'Sensor cracking' issues put me off completely. I'll wait for the M10.

 

I do not participe much on the M9 section because as I don't have a M9, my comments are not really helpful.

 

That said, I've used a few times a M9 and I see "red on the left" (too) often (even after the firmware update) and I cannot stop thinking there's a problem concerning the position of the sensor. I don't need either to own a M9 to give my opinion.

 

I know you disagree and I can understand your position.

 

Leica should give us an explanation. The important thing is how Leica is going solve the problem... with firmware or in the next M model?

 

If you don't mind using a camera with this problem, it's your choice. I do mind using a camera with this issue. Neither the "Red Contamination" and "Sensor cracking" issues are normal fonctions of a digital camera.

 

If there's a hardware problem, Leica should do a product recall. Maybe that's the reason why Leica doesn't communicate on these issues.

 

Whatever, we will see what happens with the M10. I would really like to trust Leica and I will be extremely happy to buy a new M body to replace my two M8s.

 

Regards,

 

ArtZ

[wlaidlaw]

Manuel,

 

In real life, even with lenses as wide as WATE, you only see red edge and dramatic vignetting on snowscapes and then only at 16 and 18mm. Remember a lot of people add vignetting to wide angle photos to make them look more exciting. The M9 does it automatically ;-}}

 

Other than that, I have not had a single problem with my M9. My poor M8 sits forlorn on a Visoflex III, rarely used.

 

Wilson

[jaapv]

Hi Jono,

 

You're probably right when you say "this whole theory stems from Mark's idea that it was impossible to fit a full frame sensor in an M8 body".

 

I haven't bought a M9 mainly because three reasons (in the following order):

 

1.- Too many problems (and too many trips Paris-Solms) until I got my M8s working properly. I wanted to know the problems first.

 

2.- Wanted a real chrome top (not grey paint). I decided to wait because I thought Leica would offer a chrome top.

 

3.- ´Red on the left' and 'Sensor cracking' issues put me off completely. I'll wait for the M10.

 

I do not participe much on the M9 section because as I don't have a M9, my comments are not really helpful.

 

That said, I've used a few times a M9 and I see "red on the left" (too) often (even after the firmware update) and I cannot stop thinking there's a problem concerning the position of the sensor. I don't need either to own a M9 to give my opinion.

 

I know you disagree and I can understand your position.

 

Leica should give us an explanation. The important thing is how Leica is going solve the problem... with firmware or in the next M model?

 

If you don't mind using a camera with this problem, it's your choice. I do mind using a camera with this issue. Neither the "Red Contamination" and "Sensor cracking" issues are normal fonctions of a digital camera.

 

If there's a hardware problem, Leica should do a product recall. Maybe that's the reason why Leica doesn't communicate on these issues.

 

Whatever, we will see what happens with the M10. I would really like to trust Leica and I will be extremely happy to buy a new M body to replace my two M8s.

 

Regards,

 

ArtZ

It seems a bit overdone to do a recall on far over 10.000 cameras for a component problem on less than 10. So far the M9 appears to be one of the less problematic new cameras in the industry.

[lct]

...So far the M9 appears to be one of the less problematic new cameras in the industry.

[jonoslack]

Hi Jono,

 

You're probably right when you say "this whole theory stems from Mark's idea that it was impossible to fit a full frame sensor in an M8 body".

 

I haven't bought a M9 mainly because three reasons (in the following order):

 

1.- Too many problems (and too many trips Paris-Solms) until I got my M8s working properly. I wanted to know the problems first.

 

2.- Wanted a real chrome top (not grey paint). I decided to wait because I thought Leica would offer a chrome top.

 

3.- ´Red on the left' and 'Sensor cracking' issues put me off completely. I'll wait for the M10.

 

I do not participe much on the M9 section because as I don't have a M9, my comments are not really helpful.

 

That said, I've used a few times a M9 and I see "red on the left" (too) often (even after the firmware update) and I cannot stop thinking there's a problem concerning the position of the sensor. I don't need either to own a M9 to give my opinion.

 

I know you disagree and I can understand your position.

 

I'm not disagreeing with the theory because of any loyalty, or because I don't think it's an issue; simply because that, even if the sensor were off centre I don't see that it explains the issue (see Michael's excellent posts above)

 

Leica should give us an explanation. The important thing is how Leica is going solve the problem... with firmware or in the next M model?

 

I'm not sure that they have one! I know then Kodak never had one for the issue on the SLR'n. as for solving it - for leica lenses I really really don't think it's an issue - and I wouldn't expect them to bust a gut spending money on third party lenses - so I wouldn't expect a firmware update.

 

 

If you don't mind using a camera with this problem, it's your choice. I do mind using a camera with this issue. Neither the "Red Contamination" and "Sensor cracking" issues are normal fonctions of a digital camera.

 

I can honestly say that not one shot out of something like 25,000 in more than a year has been spoiled by the 'Red Contamination' - although I do have to admit to only using Leica lenses, and not having an 18mm SE - so if you want to use 3rd party wide angles extensively, and you don't want to use cornerfix, I can see that might be a killer.

 

As for the sensor cracking - Jaap has it, it's happened rarely, and the fact that most of the examples happened close together suggests to me a materials issue.

 

Mind you - I'm not suggesting that there aren't issues, just that I don't think that these two are of any serious consequence

 

If there's a hardware problem, Leica should do a product recall. Maybe that's the reason why Leica doesn't communicate on these issues.

 

Whatever, we will see what happens with the M10. I would really like to trust Leica and I will be extremely happy to buy a new M body to replace my two M8s.

 

Regards,

 

ArtZ

If there was a hardware problem I'm sure that Leica would do a hardware recall (they did one for the M8).

[lars_bergquist]

I must say that with this discussion, I am wholly on Jonathan's side. Whatever the explanation of the rededge effect is (and how the Devil did THAT insinuate itself into this thread?) it is probably not a simplistic one. And the fact that MY sensor filter cracked, and I am of course very frustrated, does not mean that the Universe is going bust or that the M9 is a shitty camera. Apart from this problem, the M9 has been very dependable indeed and I am not going to exchange it for some mirror-flapping plastic wonder.

 

But I will be very happy when I can shift my photography back from Nokia to Leica ...

 

The old man from the Age When Shit Did Happen

[jonoslack]

HI Lars

I must say that with this discussion, I am wholly on Jonathan's side. Whatever the explanation of the rededge effect is (and how the Devil did THAT insinuate itself into this thread?)

Because of Shootist's categorical remark on message number 57 of the thread that all problems hinged around an off centre sensor . . .

 

Any idea when you'll get your camera back Lars? Did you send it straight to Solms?

[kuspen]

No it´s my turn : ( The sensorglass cracked all by itself yesterday. Finded it out this morning when shimping. It`s very hot here in Sweden. Thermal cause???

[Advertisement]

Hallo,

Look here: Leica M

[jonoslack]

No it´s my turn : ( The sensorglass cracked all by itself yesterday. Finded it out this morning when shimping. It`s very hot here in Sweden. Thermal cause???

 

Poor you . . . . I hope it gets fixed very fast . . . is it worth asking when you bought the camera?

 

all sympathies

[Baptiste]

Poor you . . . . I hope it gets fixed very fast . . . is it worth asking when you bought the camera?

 

all sympathies

 

It should be a reflex now... When reporting a sensor crack, one should as well mention the receipt date.

 

I feel for you mate, this is a puzzling issue .