Banding, some more tests

[Guest guy_mancuso]

Advertisement (gone after registration)

Has anyone entertained the thought of the micro lenses on the outside of the sensors. Is the coding on or off? is this with or without coded lenses. There are so many variables here . Does it happen at dead center of the lens or just on the outsides? Is it only with certain types of lighting. Does it go away when stoped down with the same EV value? Certain lenses? Doe' s it happen with only jpegs or raws or both and what raw converters. Only at 1250 and not 3200? i have not read all of this thread and not sure i want to, rather have the camera in my hand or some raws files to process to see.

[rvaubel]

Well, there is still a problem. I understand what is going on. The problem is that this camera cannot handle it. Other cameras can. I measured this light with a calibrated spot meter. EV12. Background EV4. I run across these sort of situations all the time. I cannot reduce the exposure as you suggest without getting into other issues such as increasing the noise in already noisy shadow areas. And making sure that there are no lights in the framing of my photos is not really a solution either. This has got to be addressed.

 

I didn't say it solved anything. You have to expose the shot correctly. By "exposure" I meant the slider in PS Raw. My little speal just explains whats going on. Your right, the blown highlights (which should be blown) are causing the banding. The question is why. Software, microlenses, sensor or what?

 

Rex

[Guest guy_mancuso]

Well blowing a highlight in any camera will cause blooming and stuff like that and if the exposure is so far off no sensor will handle it. Has anyone purposely underexposed a scene for the highlights and still getting the banding

[sean_reid]

I hadn't notice any casts in Sean's pictures either, even the indoor, incandescent lit still lifes.

 

Yes, I was thinking of those vegetable pictures in particular. Color was excellent in fact.

 

Cheers,

 

Sean

[clayh]

The exposure is correct. The sensor just does not handle the clipped highlights and the pixels downstream of them appropriately. I suspect that what we are seeing is just noise caused by the light source that is bleeding into the rows of pixels downstream of them. I doubt if the microlenses are a huge issue, since I can create the problem with the Noct, which is a slightly longer than normal focal length on the M8. All I know is that there will be some busy engineers tomorrow morning in Solms.

 

I didn't say it solved anything. You have to expose the shot correctly. By "exposure" I meant the slider in PS Raw. My little speal just explains whats going on. Your right, the blown highlights (which should be blown) are causing the banding. The question is why. Software, microlenses, sensor or what?

 

Rex

[clayh]

Yes, I have, and the banding is still there. My $200 point and shoot digital is able to handle the same situation with aplomb.

 

Well blowing a highlight in any camera will cause blooming and stuff like that and if the exposure is so far off no sensor will handle it. Has anyone purposely underexposed a scene for the highlights and still getting the banding

[jrgeoffrion]

Advertisement (gone after registration)

Has anyone entertained the thought of the micro lenses on the outside of the sensors. Is the coding on or off? is this with or without coded lenses. There are so many variables here . Does it happen at dead center of the lens or just on the outsides? Is it only with certain types of lighting. Does it go away when stoped down with the same EV value? Certain lenses? Doe' s it happen with only jpegs or raws or both and what raw converters. Only at 1250 and not 3200? i have not read all of this thread and not sure i want to, rather have the camera in my hand or some raws files to process to see.

 

Quick answers:

 

- 640, 1250, and 2500 ISO;

- On the edges for sure -- not certain about center;

- Coding was off in all cases I've seen;

- Raw (should be with JPG too but haven't seen any JPG from the M8);

- I've seen 35 'cron ASPH and 50 'noc;

- Happens with all raw converters.

[sean_reid]

I've looked at two of Pascal's DNG and the color cast is actually in the raw file itself. He posted the MacBeth chart previously somewhere (can't remember the exact thread). When the grays are white balanced, the rest shift to magenta under the halogen lights Pascal used. This did not happen with his DMR.

 

Hi JR,

 

If WB is neutral (and it is in both of the two samples Pascal showed with the WhiBal cards) then the problem is related to a color profile. As you know, the information in a RAW file can end up many different ways depending on what happens during conversion. I don't know if this could be Mac related but I'm not seeing the cast in conversions made with an XP machine. In any case, this magenta cast is not something I think people should worry too much about. If C1 needs to tweak the (Mac?) profile then it's easy enough to do that and make it available via the web.

 

I mentioned above or in another thread (who can keep track?) that I have yet to test a camera that performs auto WB accurately in all lighting conditions. None of my Canons can do that, nor can the Nikon D200. For precise color, one needs to either set WB in RAW conversion or dial it in manually. Most cameras can give decent auto WB in daylight but in artificial light they tend to be all over the map. For people working in RAW, auto WB is of little consequence. For JPEG shooters, trusting any camera's auto WB for critical work is taking a huge gamble.

 

That leaves the light streaking and I will indeed be looking at that this week as will, I'm sure, many other people.

 

Cheers,

 

Sean

[clayh]

Here is a crop from a frame where I put the offending light source in the middle of the frame, underexposed -3EV, and shot at ISO1250. Notice the band and the peculiar ghost image to the left of the light. 35mm Lux with no filter.

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!

[rvaubel]

Has anyone entertained the thought of the micro lenses on the outside of the sensors. Is the coding on or off? is this with or without coded lenses. There are so many variables here . Does it happen at dead center of the lens or just on the outsides? Is it only with certain types of lighting. Does it go away when stoped down with the same EV value? Certain lenses? Doe' s it happen with only jpegs or raws or both and what raw converters. Only at 1250 and not 3200? i have not read all of this thread and not sure i want to, rather have the camera in my hand or some raws files to process to see.

 

I'm with you. I want the camera just so I can solve the mystery! For now I'll use the RD1 for pictures;)

 

Rex

[ho_co]

Has anyone entertained the thought of the micro lenses on the outside of the sensors.

Guy--

Andy Piper mentioned earlier that it doesn't seem to be the micro lenses since they are radially arranged but the streaking is always parallel to the long axis of the frame.

 

--HC

[clayh]

Here is one with the lens coding on. 28 Elmarit ASPH. ISO 1250 minus 2 EV, light source in center of frame. Banding still occurs. Both RAW and jpeg show the banding.

 

 

 

Quick answers:

 

- 640, 1250, and 2500 ISO;

- On the edges for sure -- not certain about center;

- Coding was off in all cases I've seen;

- Raw (should be with JPG too but haven't seen any JPG from the M8);

- I've seen 35 'cron ASPH and 50 'noc;

- Happens with all raw converters.

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!

[Guest guy_mancuso]

FYI i just opened C1 with a DMR file with the leica R9 generic profile than I switched it to the M8 profile and magenta did come up stronger. so go back to the offending magenta file and chose the R9 generic in C1 and see if the magenta goes away. This maybe easy fix

[clayh]

Cool. I was never too exercised about the magenta thing. That can always be fixed in raw processing. Nice catch, though. I just went to try this and it works wonders for mixed lighting indoors. I actually preferred the R8 profile.

 

FYI i just opened C1 with a DMR file with the leica R9 generic profile than I switched it to the M8 profile and magenta did come up stronger. so go back to the offending magenta file and chose the R9 generic in C1 and see if the magenta goes away. This maybe easy fix

[jrgeoffrion]

Hi JR,

 

If WB is neutral (and it is in both of the two samples Pascal showed with the WhiBal cards) then the problem is related to a color profile. As you know, the information in a RAW file can end up many different ways depending on what happens during conversion. I don't know if this could be Mac related but I'm not seeing the cast in conversions made with an XP machine. In any case, this magenta cast is not something I think people should worry too much about. If C1 needs to tweak the (Mac?) profile then it's easy enough to do that and make it available via the web.

 

I mentioned above or in another thread (who can keep track?) that I have yet to test a camera that performs auto WB accurately in all lighting conditions. None of my Canons can do that, nor can the Nikon D200. For precise color, one needs to either set WB in RAW conversion or dial it in manually. Most cameras can give decent auto WB in daylight but in artificial light they tend to be all over the map. For people working in RAW, auto WB is of little consequence. For JPEG shooters, trusting any camera's auto WB for critical work is taking a huge gamble.

 

That leaves the light streaking and I will indeed be looking at that this week as will, I'm sure, many other people.

 

Cheers,

 

Sean

 

Sean, the cast is very specific to the light source. It's more a color shift than a color cast if I can express it better that way. I'm on a Win machine and took the raw MacBeth Pascal shot and just performed a standard WB on the grey patch. Row 7 of the DCC (brownish skin tones) shifted to magenta (as well as the reds and yellows). On the other hand, the same DCC under the same lighting with the DMR doesn't have this color shift. This is, IMHO, like you mentioned, more of a color profile tweaking than anything else -- I wouldn't call this an issue.

[Guest guy_mancuso]

Thanks Clay for posting these . So the raw converter has no effect here and center or on the outside has no effect . Also coded or not. Only at 640 and above. Some people have it and some don't . This is starting to sound like a bad sensor. Have to look at more of this, interesting it is happening only at the high ISO which at these settings the sensor is more sensitive to light because it is trying to gather more info and so the light charge goes up. Not a engineer here folks but the charge is also horzontal and no one has seen any vertical banding.

 

Howard thanks just read that as I am writing this section.

 

 

Reason part of this is do to a very sensitive sensor to begin with because it is capturing such a wide dynamic range reason the P&S camera's don't do this BTW because the sensor are not even this close to the sensitivty level of the Pro sensors( Even the Canons are about 2 stops short of the M8 and DMR in Dynamic range). They have maybe a 7 stop range and these more like 10 or 11. Those numbers vary but hope the point is there.

 

I am starting to think this is a light charge issue as it travels across the horizontal plane. So that theory is very good thinking but why some sensors and not all, is the part that is the question. Bad firmware or bad sensor. Having a hard time being a bad sensor since the DMR is in the same family here

[clayh]

Guy, if you look closely at the ISO test I posted a page or two back, you can see the faint glimmerings of the banding beginning to appear even at ISO 160. From what I have read, high ISO setting are really accomplished by setting the 'gain' higher for the sensor. And just like when you crank the volume dial on your stereo receiver up, you get more noise, and in this case, I am guessing it also amplifies the noise along that pixel 'path' caused by the over-amped pixels where the light source is located. There might possibly be some constructive interference (feedback) also occuring along that channel.

 

 

 

Thanks Clay for posting these . So the raw converter has no effect here and center or on the outside has no effect . Also coded or not. Only at 640 and above. Some people have it and some don't . This is starting to sound like a bad sensor. Have to look at more of this, interesting it is happening only at the high ISO which at these settings the sensor is more sensitive to light because it is trying to gather more info and so the light charge goes up. Not a engineer here folks but the charge is also horzontal and no one has seen any vertical banding.

 

Howard thanks just read that as I am writing this section.

 

 

Reason part of this is do to a very sensitive sensor to begin with becuase it is capturing such a wide dynamic range reason the P&s camera's don't do this BTW becuase the sensor are not even this close to the sensitivty level of the Pro sensors. They have maybe a 7 stop range and these more like 10 or 11. Those numbers vary but hope the point is there.

 

I am starting to think this is a light charge issue as it travels across the horizontal plane. So that theory is very good thinking but why some sensors and not all, is the part that is baffiling.

[pascal_meheut]

Color cast is not an OS issue: I tried both on Mac OsX (last version) and Windows (W2K SP4) and I got the exact same file out of Capture One 3.7.6. Well, there are some differences because C1 does not expose exactly the same and so on both platforms but here is the comparison.

 

I'm pretty sure this is a profile issue and we need to ask PhaseOne to build some more profiles for the M8 under different light conditions, just like they do for the Canon Eos1DsII, their P25 backs and so on.

I'm not a registered user because Leica forgot the sticker on my C1 LE CD so I'm using the trial version and a friend's license but I'll write to them as soon as possible if Leica does not come back to me on this one.

And yes, I do not consider this a major issue, just a minor annoyance, very easy to fix. I just would like to have some PhaseOne profiles because the ones I'm building myself are up to their quality yet

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!

[J_Brittenson]

The M8 issue looks like it falls in the D200/1Ds category - signal contaminating the downstream pixels.

Well, we can all speculate over the cause. My speculation would be that a severely saturated signal at high gain (ISO) causes charge buildup and a memory effect, either in the CCD readout amplifier, or the ADC's input amplifier. Further, I'd speculate that the streak is really gray (neutral across channels), but becomes magenta after WB adjustment for very green (fluorescent) light. Hence, if you shoot ISO 2500 in daylight it'll show as a luminance variation, and actually may not show at all because of the less strong WB correction (relative levels shifting). If my hypotheses are correct, then the workaround for now might be to pull out the old fluorescent-to-daylight filters and use until there's a fix, to confine the effect to luminance, or to reduce it to the invisible entirely.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[pascal_meheut]

BTW, just got an answer from Leica, first thing in the morning. These guys are responsive for sure.

They are aware of the problem and working on a firmware fix for the moment.