Dead pixel? on a week old Leica M

[TomB_tx]

Advertisement (gone after registration)

What's the prognosis if the fault appears to be intermittent? I've just looked through some files from yesterday and have been shocked to see the same thing, single vertical line. Started late afternoon, early evening.

 

Just tried a few shots today, no sign of it at all. Tried replicating settings, lenses, used the same card, all files were perfect. Any ideas??

I believe sensor pixels can be "mostly dead" so as to show up sometimes and not others. Perhaps it ties into sensor temperature, so the problem shows up towards the end of an active day, but cools and somewhat recovers overnight. The M allows live-view which does heat the sensor more than you would have with an M9. Perhaps that is why there seem to be more such problems on the M.

[Advertisement]

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

[dave.gt]

Every Nikon DSLR I have ever owned had multiple dead pixels. D1...D2H...D2X... And a D40. None on the X1. Purely random chance IMO.

[Daniel Beacock]

I believe sensor pixels can be "mostly dead" so as to show up sometimes and not others. Perhaps it ties into sensor temperature, so the problem shows up towards the end of an active day, but cools and somewhat recovers overnight. The M allows live-view which does heat the sensor more than you would have with an M9. Perhaps that is why there seem to be more such problems on the M.

 

Interesting point. And actually that day was the first day I had used the EVF for the majority of the time trying out some old Olympus OM and Canon FD lenses. I'll try again this weekend and see if I can replicate it.

[Urmelchen]

Dear Christianpsu,

 

This is the reality in digital imaging sensors, Leica and all others.

 

This is an economic question, and a defect free sensor is not a commersial product, but is enjoyed by military and space customer, at an similarly astronomic price.

 

The spec of the M8 (check the Kodak datasheet for the KAF-10500 sensor ) allows for 2500 pixel defects, but less than 30 in a cluster, but up to 10 dead columns.

 

Trade spec on full format is less than 10 defect columns for full (24X36) format, and less than 20 columns in medium format.

 

It seldom gets that bad, but you cannot complain if you get a camera with that many defects.

 

This is standard consumer grade, and the same for Canon, Nikon and others.

 

My M9P has three dead columns, my M M had two completely dead columns, so had my M type 240. My Phase One IQ180 digital back has one half dead column, and several clusters of dead pixels.

 

My Canon 5D2 and Nikon D800E has no dead columns, but scattered dead pixel clusters all over the place.

 

I complained to Leica on the three dead columns on my M9P, but they claim that the sensor is within spec.

 

This is just how the state is in 2013, no (almost) sensors are perfect, this is the reality in microelectronics manufacturing.

 

I am and IC designer, and have worked on image sensors, this is just the reality, ant not a big problem for image quality as these defects are efecctively mapped out.

 

The only occasion when these defects are visible are with motifs with repetitive patterns, like roof tiles or building facades with window blinds etc.

 

One very easy way to identify these defects is to take your sharpest lens APO75, APO50 or SX50 etc at f8, and take a photo of a 24 inch computer monitor in focus at a distance of approx 70cm - 1 meter, of an all white image.

 

This will give an image at approx 1/4-1/8 the Nyquist frequency of the sensor, and every known defect correction will go absolutely haywire, and you will immedeately see the dead columns and clusters.

 

Best regards

 

Trond

 

Have done this test and shoot my IMac 27 screen.

I'm getting 7 bad lines!

 

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!

[pico]

Have done this test and shoot my IMac 27 screen.

I'm getting 7 bad lines!

 

Could you make a picture from a more natural object? Making a photo of a small space of a CRT is the worst way to examine sensor performance. If you need an explanation, it will be forthcoming but the issue should be obvious.

 

So can you do that for us?

[zlatkob]

My Canon 5D2 and Nikon D800E has no dead columns, but scattered dead pixel clusters all over the place.

 

I've seen a few hot pixels with Canon cameras, but only with the medium & small Raw files. This is never a problem with the full Raw file size because Lightroom automatically removes any hot pixels in large Raw files — a wonderful feature built into Lightroom. However, I believe it only fixes individual pixels, not a whole column.

[Urmelchen]

Advertisement (gone after registration)

Could you make a picture from a more natural object? Making a photo of a small space of a CRT is the worst way to examine sensor performance. If you need an explanation, it will be forthcoming but the issue should be obvious.

 

So can you do that for us?

 

On normal photos the lines are not visible. Somtimes the third row from the left is visible in shadow areas. But not on every picture.

 

Would be nice if other M240 user would do the test too.

[StephenPatterson]

I seem to have one mapped out column on the extreme right side. This is with my M240 that shipped new from the factory approx. one month ago.

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!

[StephenPatterson]

Could you make a picture from a more natural object? Making a photo of a small space of a CRT is the worst way to examine sensor performance. If you need an explanation, it will be forthcoming but the issue should be obvious.

 

So can you do that for us?

 

The only thing this test shows is how many columns or possibly pixel clusters have been mapped out. In the real world the sensor mapping algorithm will camouflage these areas quite well. As others have pointed out dead pixels and columns in digital sensors are normal, but having to send the camera back to the mother ship to correct this in software is frustrating to say the least.

[MarkP]

Dear Christianpsu,

 

This is the reality in digital imaging sensors, Leica and all others.

 

This is an economic question, and a defect free sensor is not a commersial product, but is enjoyed by military and space customer, at an similarly astronomic price.

 

The spec of the M8 (check the Kodak datasheet for the KAF-10500 sensor ) allows for 2500 pixel defects, but less than 30 in a cluster, but up to 10 dead columns.

 

Trade spec on full format is less than 10 defect columns for full (24X36) format, and less than 20 columns in medium format.

 

It seldom gets that bad, but you cannot complain if you get a camera with that many defects.

 

This is standard consumer grade, and the same for Canon, Nikon and others.

 

My M9P has three dead columns, my M M had two completely dead columns, so had my M type 240. My Phase One IQ180 digital back has one half dead column, and several clusters of dead pixels.

 

My Canon 5D2 and Nikon D800E has no dead columns, but scattered dead pixel clusters all over the place.

 

I complained to Leica on the three dead columns on my M9P, but they claim that the sensor is within spec.

 

This is just how the state is in 2013, no (almost) sensors are perfect, this is the reality in microelectronics manufacturing.

 

I am and IC designer, and have worked on image sensors, this is just the reality, ant not a big problem for image quality as these defects are efecctively mapped out.

 

The only occasion when these defects are visible are with motifs with repetitive patterns, like roof tiles or building facades with window blinds etc.

 

One very easy way to identify these defects is to take your sharpest lens APO75, APO50 or SX50 etc at f8, and take a photo of a 24 inch computer monitor in focus at a distance of approx 70cm - 1 meter, of an all white image.

 

This will give an image at approx 1/4-1/8 the Nyquist frequency of the sensor, and every known defect correction will go absolutely haywire, and you will immedeately see the dead columns and clusters.

 

Best regards

 

Trond

 

Hi Trond,

thanks for this very clear explanation.

However I don't perceive any dead-pixels in my photos so I'd dread doing this test. If I found any faults I wouldn't sleep at night.

Ignorance is bliss .

 

Regards,

Mark

[trond]

Dear Mark,

 

You will not be able to see dead pixels with this test, as they are mapped out and corrected for.

 

However, you will be able to see dead columns even if they are corrected, because the correction can not handle the repeated pattern of the LCD screen.

 

I have had three different M9, and one M M, and they all had two to three dead columns. My current M9-P does not have any column defects.

 

Even my Phase One IQ180 digital back, has one dead column.

 

My point was, that this is they fact of life with all digital cameras, and the a defect free sensor very seldom happens.

 

There are always several pixel defects, standard consumer grade accepts a few thousand, and even several dead columns, up to 10 in 24x36 senors and up to 20 in medium format.

 

These defects are easy to spot on certain image motifs, specially in architecture photos and photos of wowed textiles and fabrics.

 

Best regards

 

Trond

[trond]

Have done this test and shoot my IMac 27 screen.

I'm getting 7 bad lines!

 

[ATTACH]392030[/ATTACH]

 

Dear Urmelchen,

 

This is a good example of what can be expected.

 

Try to load the image into Photoshop and make a note of the column number of the defects.

 

Then see if you are able to se them in real images.

 

I expect that you will not be able to do so, except for images of rooftiles, typically cityscape, or woven fabrics.

 

Best regards

 

Trond

[Urmelchen]

Dear Urmelchen,

 

This is a good example of what can be expected.

 

Try to load the image into Photoshop and make a note of the column number of the defects.

 

Then see if you are able to se them in real images.

 

I expect that you will not be able to do so, except for images of rooftiles, typically cityscape, or woven fabrics.

 

Best regards

 

Trond

 

Hi Trond,

 

sometimes i see the third line frome left in dark areas as a line.

Should i call Leica customer support about this? 7 lines are pretty much for such a expensive camera. My Canon 5D2 don't have any bad line.

[StephenPatterson]

Hi Trond,

 

sometimes i see the third line frome left in dark areas as a line.

Should i call Leica customer support about this? 7 lines are pretty much for such a expensive camera. My Canon 5D2 don't have any bad line.

 

Lines are not as bad as clusters, as there is usually enough surrounding information for the algorithm to select a very close value to what would have actually been recorded if the pixels were active.

[trond]

Hi Trond,

 

sometimes i see the third line frome left in dark areas as a line.

Should i call Leica customer support about this? 7 lines are pretty much for such a expensive camera. My Canon 5D2 don't have any bad line.

 

Dear Urmelchen,

 

I do not think this is any reason for complaint.

 

Camera manufacturers do not publish what is acceptable defect levels.

 

However, you can find what can be expected by looking at sensor data sheets from manufacturers, Kodak, Truesense, Dalsa, Aptina and the like.

 

If you look at page 16 of the Kodak KAF-10500 sensor used in the M8, you find a very good listing of what can be expected.

 

Please note that this is what is called "standard grade", and that camera manufacturers may apply stricter criteria in their sensor purchasing.

 

It is all about the price, a "perfect" defect free sensor for a 24x36mm camera will easily cost several thousand dollars for the sensor alone.

 

Best regards

 

Trond

[Urmelchen]

Dear Urmelchen,

 

I do not think this is any reason for complaint.

 

Camera manufacturers do not publish what is acceptable defect levels.

 

However, you can find what can be expected by looking at sensor data sheets from manufacturers, Kodak, Truesense, Dalsa, Aptina and the like.

 

If you look at page 16 of the Kodak KAF-10500 sensor used in the M8, you find a very good listing of what can be expected.

 

Please note that this is what is called "standard grade", and that camera manufacturers may apply stricter criteria in their sensor purchasing.

 

It is all about the price, a "perfect" defect free sensor for a 24x36mm camera will easily cost several thousand dollars for the sensor alone.

 

Best regards

 

Trond

 

Sometimes it's looking like this. Not very good for a 6200 Euro 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!

[trond]

Dear Urmelchen,

 

I agree with you on that, this should not be visible in this kind of image.

 

It looks to me like the the pixel remapping is not working correctly, since you are getting a blue line.

 

Which RAW converter did you use?

 

My Phase One IQ180 shows a similar blue line at the same position in every image, but only visible if I use Lightroom.

 

The blue line is not visible in other raw converters.

 

Best regards

 

Trond

[Urmelchen]

Dear Urmelchen,

 

I agree with you on that, this should not be visible in this kind of image.

 

It looks to me like the the pixel remapping is not working correctly, since you are getting a blue line.

 

Which RAW converter did you use?

 

My Phase One IQ180 shows a similar blue line at the same position in every image, but only visible if I use Lightroom.

 

The blue line is not visible in other raw converters.

 

Best regards

 

Trond

 

Hi Trond,

 

I've tried all common Raw Converters (LR5,C1,Aperture,DXO,RawTherapee). It's visible with all more or less.But only sometimes not in all pictures.

[Advertisement]

Hallo,

Look here: Leica M

[Guest WPalank]

William, I tried to post the full image, but with the size rules there was no way you can see the line when you zoom in, it is simply too pixelated to be defined at all. So here are two crops. Looking at it closer in this image now I realise it's not a continuous line, it does appear to be broken, particularly in lighter areas. Again, today the camera is absolutely fine, still no further sign of the problem. In these crops the line should be evident roughly in the centre, on the full image the lines appeared pretty much at the beginning of the last third of the width of the image.

 

 

Thank you Daniel. I was hoping to get an image which you think "Does not have the line" as well, to see if I can tease it out.

[christianpsu]

Well, there are dozens of threads on the subject here and in other camera forums and plenty of scientific -and less scientific - papers on the Internet. A dead pixel is a pure matter of chance, as it is caused by cosmic radiation. As the sensor s read out in columns it can show up as a line. A sensor is more at risk during flights, especially transpolar ones, due to the higher intensity of the radiation. Sony is said to ship their cameras by land/sea for this reason.

Leica had a warning in the M8 and M9 manuals. I have not checked, but it is probably still there in the M manual.

 

The only risk-reducing strategy I have come up with is target reduction, i.e. store the camera with the sensor vertical during flight. It is no scientific proof, but I have never had a dead pixel since I went digital in 1998. despite quite an amount of accumulated flying time -touch wood.

 

Any camera grade sensor has dead pixels from new and they are mapped out (i.e. assigned a value based on the neighbouring pixels) before the camera is shipped.

 

Once a dead pixel appears on the sensor during use it must be remapped. Some cameras have a sensor remapping feature in the firmware, but Leica prefers to do the routine in the factory. There is a maximum number of pixels that can be mapped out and when that tolerance is exceeded the sensor will be replaced.

 

All in all a nuisance but a fact of digital life.

 

I will try this vertical storing method while I am on board a plane. I mean if you need a 8 feet thick concrete wall to block out cosmic rays while in the sky, then I don't think my Billingham bag will be any helpful (including the steel plates that the plane is made of). Thanks jaapv.