Diffusion of red.

[tonybeach]

Michael, the photo is not "bleeding", it was shot a year and half ago and is actually slightly underexposed (-.3 EC, before I knew what I was doing), sunlight is illuminating the leaves from behind and I think you are mistaking out of focus leaves blurring into one another as some sort of problem with the sensor. Most of my flower shots are flash or have flash fill but I was able to dig up an old D70 shot: http://photos.imageevent.com/tonybeach/myfavorites/_DSC2380_websized.jpg<div></div>

[mmene]

What is this?

[bjørn rørslett]

I'd say the post processing.

[tonybeach]

Thank you Bjorn, in fact it is an example of the clumsy PP I was doing a year and a half ago -- I like to think that I get both better exposures and handle them better now.<div></div>

[mmene]

Thanks to all of you my friends. I think Bjorn is right.

[mmene]

About red oversaturation.

Let's consider light red color (pink). In a case of light red, R has a larger value than that of G and B. For example, R=200, G=100, and B=100.

At this time, the values of G and B are a half of the R value. When this ratio is maintained, the color looks the same.

For example, in a case of darkened light red, the values are R=100, G=50 and B=50. When it makes brighter, the values become R=200, G=100, and B=100.

When it makes more brighter, for example, in a case of the most brightest light red, the values become R=254, G=127, and B=127. (*2)

From this condition, if adjusting the color much more brighter, what will happen? On paper, the values should become R=300, G=150, and B=150.

However, since R=300 cannot be represented in reality, then the actual value of brightness becomes R=255. As the result, the three values are R=255, G=150 and B=150 in this case.

 

What about setting the color much more brighter? The values R=510, G=255 and B=255 become R=255, G=255 and B=255. This is pure white color!

Roughly speaking, the "over-saturation" occurs in this way.

 

In the case of this light red color, the limit values that the hue does not change are R=254, G=127 and B=127. And the brightness that human eyes can sense is 254+127+127=508.

This means that although the brightness of white can be represented up to 765, the brightness of light red can be represented only up to 508.

 

How about a case of deeper red color? For example, let's consider a case of deeper red with values as R=200, G=50 and B=50.

The values of the most brightest deeper red that can be represented are R=252, G=63 and B=63. This means that its maximum brightness is 252+63+63=378.

 

Therefore, when a color is deeper and deeper, brighter representation cannot be possible.

So when an exposure bias is adjusted to a light color subject, the "over-saturation" occurs at a deep color subject.

This is the reason for the "over-saturation" occurrence.

 

That's the anser my friends.

[tonybeach]

Micheal, you do not know what you're talking about. The red channel on the sensor captures fewer photons than the green channel. If the green channel is not overexposed, the red channel is not overexposed. How you choose to handle that data by applying the WB coefficient and what color space you convert the file to are what causes the red channel to clip.

[mmene]

I know very well what I am talking about but there is no more discussion with folks of this style. It's not the style I like to my companions. So I wish you the best my friend and peace on earth to everybody .

[mmene]

About OVERSATURATION OF RED. http://www.xs4all.nl/~tindeman/raw/color_reproduction.html

[tonybeach]

Michael, you do not understand the role of WB gain in the exposure of the red channel. Over-saturation occurs after the red channel has been boosted and not before, if you had read and understood the link to the previous thread where I discussed this than you would realize you are wrong. I provided a simple experiment you can use to test what I'm saying for yourself -- again, test and then get back to me with your results.