What happened with the red color here?

[dennersten]

Advertisement (gone after registration)

What happened here with the reds. Seems to be some strange glowing color in the out of focus areas. It only happens to me with some flowers. Never seen it in other situations. Are there some solutions, seems to be very tricky to fix afterwards?

 

Picture are taken with M240 and Apo 50 (but i've seen the same problem with other lenses), no processing.

The text means "old joinery" in Swedish.

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!

[dorman]

The red channel is saturated.If you shot in dng you may be able to recover it.

I don't know much about what control the M240 gives you, but for better jpegs in this type of situation you could try:

Reduce the Saturation setting.

Reduce the Contrast setting.

Reduce the exposure.

Edited July 12, 2014 by dorman
Added more information

[Advertisement]

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

[pfhrased]

CMOS sensor bloom. Red channel is prone to it on a variety of different cameras.

[thighslapper]

The M habitually over-exposes images with large amounts of red..... (as the sensor is least sensitive to red ..... and most to green if my memory serves me correct .....I'm sure Olaf/Michael will put me right ... )

 

Several threads about this in the past ...... and a fair amount of discussion by my fellow beta testers prior to the last firmware update that corrected most of the original colour issues.

 

Under-expose by -1 ev or more and it will compensate. Trying to correct it after is more of a problem.

Edited July 12, 2014 by thighslapper

[mjh]

This is overexposure of the red channel – a fairly common phenomenon with subjects with saturated primary colours. Correcting the exposure should help.

[albireo_double]

Overexposed in the red channel, colour out of gamut. Check the histogram for the red channel. Red flowers need to be relatively under-exposed to avoid this.

[CheshireCat]

Advertisement (gone after registration)

If OP is shooting JPG, then switching to DNG will help mitigating this issue.

[Aliturk]

I'm only just back in the Leica fold, but had this problem with the D800 - and found many flower colours over staturate easily. Typical of CMOS sensors. So I used to set exposure bias to at least minus 1 and 1/2 stops, and then bring the exposure back up in lightroom to a point before losing detail in the petals. I've not tried this with my M240 yet, but I'm sure the technique will work just as well. You must shoot raw/DNG to be able to do this.

[jaapv]

I am not sure whether the sensor type has anything to do with it, but cameras do have an RGB histogram, so it is not hard to see if a colour channel is overexposed.

[mjh]

With CCDs this happens just as often as with CMOS sensors.

[Paul J]

Film does it too albeit more forgivingly. One of the benefits of digital and histograms is being able to see it.

[CheshireCat]

With CCDs this happens just as often as with CMOS sensors.

 

It actually depends on the dynamic range. The higher the dynamic range, the less likely this problem will happen.

The M9's CCD has about 2-stops lower dynamic range than the M's CMOS, so I would expect this issue to happen more often on the M9.

[mjh]

It actually depends on the dynamic range.

The problem isn’t so much the limits of dynamic range but the RGB colour space. In any RGB colour space the brightest yellow, cyan, and magenta is brighter than the brightest red, green, or blue (and of course the brightest white is even brighter). If you try to capture red and yellow of equal brightness you will either underexpose yellow or you get clipped reds.

 

If automatic exposure did prevent the overexposure of even a single colour channel then the resulting image would look quite dark whenever there were highly saturated reds, greens, or blues in the picture. But then the M’s classic metering cannot evaluate different parts of the image or even differentiate between colours anyway. Matrix metering with the image sensor could help with automatic exposure – in theory at least; I’m not sure whether it does.

[CheshireCat]

The problem isn’t so much the limits of dynamic range but the RGB colour space.

 

Uhm...

If that would be the case, then the RAW DNG would contain the correct (not saturated) color information. If the RAW contained the correct information then the issue could be fixed in post, before gamma compression and colorspace conversion.

 

Here we are talking instead of a saturated sensor (although partially, e.g. red pixels only) and sensor saturation can be fixed with a higher dynamic range... or underexposure, of course.

 

It is also important to note that, in general, highlights dynamic range decreases by increasing ISO, so it is advisable to set the camera to the lowest native ISO for this kind of shots.

Edited July 17, 2014 by CheshireCat

[jaapv]

I think it is more clear to say (the same thing) that the red channel is overexposed.

Highlight clipping is not really fixed by having more dynamic range, as dynamic range has two ends, where the highlights are overexposed abruptly as the pixel is saturated to 100%, whereas the shadows peter off into the noise floor. So it is equally possible to clip the highlights on a high-DR sensor as on low-DR sensor, you just lose range in the shadows.

Edited July 17, 2014 by jaapv

[mjh]

Uhm...

If that would be the case, then the RAW DNG would contain the correct (not saturated) color information. If the RAW contained the correct information then the issue could be fixed in post, before gamma compression and colorspace conversion.

Overexposure occurs in the sensor and affects the raw data. While one may be able to salvage some of the lost highlights when developing the original raw data, more often than not that won’t suffice.

 

As I had explained, the problem is inherent in the RGB colour space (any RGB colour space). And since the sensor pixels are sensitive to either red, green, or blue, that is where the damage is done (for example, the sensor cannot at the same time capture a bright saturated red and an equally bright saturated yellow, as either the red will be clipped or the yellow will be underexposed). Everything else – like the conversion from the device-dependent RGB colour space of the sensor to a standard RGB colour space such as Adobe RGB or gamma compression – is secondary, even when it may add to the damage. One must take care that none of RGB channels are overexposed, even when that means that the image on the whole looks underexposed.

[jaapv]

Or use a colour filter - but that will create other insoluble problems in a colour workflow

[CheshireCat]

As I had explained, the problem is inherent in the RGB colour space (any RGB colour space).

 

A CCD or CMOS sensor is monochrome. The fact a "color sensor" has a RGB bayer filter on top of it only means that a red flower will only saturate R pixels, just because G and B pixels filter red frequencies out.

 

If you take that photo without the RGB bayer matrix (e.g. M Monochrome), the red flowers would be solid white, so you would have to underexpose as well.

If you have to underexpose, then the problem is just the sensor dynamic range, not the RGB color space.

 

It is as simple as that.

 

 

for example, the sensor cannot at the same time capture a bright saturated red and an equally bright saturated yellow, as either the red will be clipped or the yellow will be underexposed

 

Why not ? The red object will be R=100, G=0, B=0; and the yellow object will be R=50, G=50, B=0 (I am assuming the eye is equally sensitive to R G and B, to keep the math simple). The yellow will have R and G "underexposed" by one stop, but the extra noise is compensated by having two samples instead of one.

Sure you'll lose spatial resolution, but that's the trade-off of the bayer color matrix.

[Exodies]

for example, the sensor cannot at the same time capture a bright saturated red and an equally bright saturated yellow, as either the red will be clipped or the yellow will be underexposed

 

Why not ? The red object will be R=100, G=0, B=0; and the yellow object will be R=50, G=50, B=0 (I am assuming the eye is equally sensitive to R G and B, to keep the math simple). The yellow will have R and G "underexposed" by one stop, but the extra noise is compensated by having two samples instead of one.

Sure you'll lose spatial resolution, but that's the trade-off of the bayer color matrix.

 

You don't seem to be disagreeing with mjh; as far as I can tell you have just provided the "or" example for his either/or.

The "either" would be the red object R=200, G=0, B=0 and the yellow object R=100, G=100, B=0.

 

This gives "either" red clipped, "or" yellow underexposed.

[Advertisement]

Hallo,

Look here: Leica M

[lethbrp]

Red sits quite low on the RGB scale, so if you have a lot of red in the image, you will need to apply exposure compensation. This is the same for any camera, I routinely do this with my Nikon.