Ugly Hedgehog - Photography Forum
Home Active Topics Newest Pictures Search Login Register
Main Photography Discussion
Why do Hi ISO pictures lose saturation of color?
Page 1 of 5 next> last>>
Mar 26, 2017 11:13:25   #
downing Loc: Cincinnati
 
Why do Hi ISO pictures lose saturation of color?

Is it a physical limitation of the sensor?

Does the reflectance of objects change with changes in Illumination level?

???

If it is the sensor what trade offs could be made to improve performance?

Reply
Mar 26, 2017 11:30:29   #
rmalarz Loc: Tempe, Arizona
 
If one is using High ISO because of very low illumination, that low illumination will result in less color vibrancy, as there is less color information available. Additionally, high ISO induces more digital grain, electrical engineers refer to this as noise. Noise degrades both color and contrast. The amount of information contained in the image is defined by the signal to noise ratio. If the signal dips towards the noise level, only the noise is discernable in the image. A low signal to noise ratio (low exposure) means less information about the colors. This results in less saturated colors.
--Bob


downing wrote:
Why do Hi ISO pictures lose saturation of color?

Is it a physical limitation of the sensor?

Does the reflectance of objects change with changes in Illumination level?

???

If it is the sensor what trade offs could be made to improve performance?

Reply
Mar 26, 2017 11:44:49   #
rjaywallace Loc: Wisconsin
 
Speaking as an old retired fellow in his 70s, I know that I don't see or read well in low-light situations. Introducing a bit more illumination tends to add color and contrast which allow me to see considerably better. Not surprising to me that it may be similar for sensors.

Reply
 
 
Mar 26, 2017 11:50:26   #
oldtigger Loc: Roanoke Virginia-USA
 
i think malarz nails it pretty well

Reply
Mar 26, 2017 12:18:18   #
Apaflo Loc: Anchorage, Alaska
 
downing wrote:
... If it is the sensor what trade offs could be made to improve performance?

There is no "grain" in a digital image, unlike with film where the density is a product of silver grains.

The problem with saturation at higher ISO values is Signal to Noise ratio. With lower light levels the way to get the same brightness in an image is to amplify the signal up to the same level as produced by higher light levels. Of course that also amplifies noise. When the signal is originally high there is a large ratio of signal to noise, but when the signal is low and is amplified to the same level there is a lower ratio of signal to noise.

The SNR also determines what is called "Dynamic Range". That is the number of fstops between the darkest tones and the brightest tones.

When the difference is about 5 or 6 fstops worth of light, setting the black level can pretty much eliminate noise visibility in a print. At maybe 7 fstops it can be invisible in an on screen image too. Note that a JPEG encoded image can only have just over 8 or maybe 9 fstops of dynamic range (because it was intensionally designed to be just better than needed!).

Essentially the answer to why saturation might be low is the way you have configured your camera, or whatever JPEG generation software you use. It need not be lower in saturation! What it absolutely is going to be, as the ISO increases, is higher in contrast. That normally would also provide saturation too. Note that "saturation" is a measure of how pure a color is. If noise is added, as might happen with extremely high ISO values (high enough to produce less than maybe 5 fstops of dynamic range) the noise will add to the signal and dilute the purity of the colors. Hence with modern cameras like the Nikon D5, which produces a dynamic range of about 4.7 fstops at an ISO of 25,600 there will be decreasing saturation (if all other adjustments are untouched) as the ISO is increased. In fact though, reseting the black point and adjusting the JPEG generator for higher saturation can easily make up the differences well into an ISO range where saturation is the least of problems!

Reply
Mar 27, 2017 06:29:02   #
cthahn
 
Give all the information. What do you mean by Hi resolution?

Reply
Mar 27, 2017 07:02:07   #
Apaflo Loc: Anchorage, Alaska
 
cthahn wrote:
Give all the information. What do you mean by Hi resolution?

You are the first to mention resolution in this thread. What do you mean?

Reply
 
 
Mar 27, 2017 07:59:56   #
jerryc41 Loc: Catskill Mts of NY
 
rmalarz wrote:
If one is using High ISO because of very low illumination, that low illumination will result in less color vibrancy, as there is less color information available. Additionally, high ISO induces more digital grain, electrical engineers refer to this as noise. Noise degrades both color and contrast. The amount of information contained in the image is defined by the signal to noise ratio. If the signal dips towards the noise level, only the noise is discernable in the image. A low signal to noise ratio (low exposure) means less information about the colors. This results in less saturated colors.
--Bob
If one is using High ISO because of very low illum... (show quote)


Right! We don't see color with enough light.

Reply
Mar 27, 2017 10:10:57   #
Fotoartist Loc: Detroit, Michigan
 
Great question! Wondering what percent of this loss of color and contrast can be made up in post processing.

Reply
Mar 27, 2017 10:17:13   #
Pablo8 Loc: Nottingham UK.
 
cthahn wrote:
Give all the information. What do you mean by Hi resolution?


Who mentioned Hi resolution??? Whose posting are you referring to??? Yet again!!!

Reply
Mar 27, 2017 10:25:14   #
selmslie Loc: Fernandina Beach, FL, USA
 
downing wrote:
Why do Hi ISO pictures lose saturation of color?

Is it a physical limitation of the sensor?

Does the reflectance of objects change with changes in Illumination level?

???

If it is the sensor what trade offs could be made to improve performance?

It has nothing to do with ISO, dynamic range or noise. It is simply the result of how the colors get mixed in the visible 8- or 16-bit image.

Take an image with a pure saturated red area. The color in the 8-bit image would be red=255, blue=0 and green=0. You can't make the red any more intense or brighter because it is at the upper limit. You can only make it darker by reducing the 255 value. As the value goes down, it initially looks pretty saturated but as it get darker (blacker) the redness declines until it is no longer clearly visible. It will have become blacker and when it reaches 0 you will see pure black.

Now try to increase the brightness. You can't add any more red. You can only add blue and green. This makes it brighter but the red saturation declines. Once the blue and green reach 255 you have white.

This happens if you start with any of the RGB colors:

Saturated Darker Lighter
Red 255,0,0 127,0,0 255,127,127
Green 0,255,0 0,127,0 127,255,127
Blue 0,0,255 0,0,127 127,127,255

If you use one of the complimentary CMY colors the numbers are a different but the logic is the same:

Saturated
Cyan 0,255,255
Magenta 255,0,255
Yellow 255,255,0

It gets a little more complicated with colors other than RGB or CMY.

The only time that high ISO gets involved is when it leads you to overexpose one or more of the primary RGB values in the raw file.

Reply
 
 
Mar 27, 2017 10:44:28   #
Apaflo Loc: Anchorage, Alaska
 
jerryc41 wrote:
Right! We don't see color with enough light.

Our eyes use different sensors for low levels of illumination than for higher levels... But the camera uses exactly the same sensors. What we see and what a camera will see are two very different things.

We might also note that "color vibrancy" has no meaning. Vibrancy is a made up term from Adobe that has no exact quality or quantity. What their vibrancy slider does changes across the range of adjustments. Changing by +10 low on the scale may do one thing, at mid scale another, and something entirely different at the high end. Part of that might be to change saturation, but another part is to change gamma and perhaps includes sharpening. It is a composite, and of unknown meaning. The point is that we cannot say an scene has more or less vibrancy depending on illumination.

"Color saturation" on the other hand is a precise technical term that means the same thing in every editor and every scene. It is how pure a color is. For example a red color represented by RGB hex values 0xE0, 0x20, and 0x30 is not pure red because the green and blue values are higher than 0. If either the red is increased, say to 0xF0, or if the green or blue are decreased, say to 0x10 and 0x20, the red color will be more saturated. Likewise if over a given area we change all values with more difference between red and either green or blue than 0xE0, 0x20 and 0x30 to values with the same differences as 0xF0, 0x10, and 0x20 and leave those already closer to the more saturated color as they are, the entire area will appear more saturated.

Reply
Mar 27, 2017 10:58:02   #
oldtigger Loc: Roanoke Virginia-USA
 
Lets rephrase the question:
If a scene is dim and you need more exposure, which will give better saturation, more ISO or slower shutter.
Or will it make any difference?

Reply
Mar 27, 2017 11:09:24   #
selmslie Loc: Fernandina Beach, FL, USA
 
oldtigger wrote:
Lets rephrase the question:
If a scene is dim and you need more exposure, which will give better saturation, more ISO or slower shutter.
Or will it make any difference?

Either one will make the image brighter. Just don't blow the highlights.

Reply
Mar 27, 2017 11:11:14   #
Apaflo Loc: Anchorage, Alaska
 
selmslie wrote:
It has nothing to do with ISO, dynamic range or noise. It is simply the result of how the colors get mixed in the visible 8- or 16-bit image.

True for ISO and dynamic range, but noise definitely affects how pure a color is, and that is what
saturation means.

Your chart is not correctly labeled:

" This happens if you start with any of the RGB colors:

Saturated Darker Lighter
Red 255,0,0 127,0,0 255,127,127
..."

The second column is darker, but just as saturated as the first column. The third column is not as saturated as the other two, but is no lighter than the first column.

"It gets a little more complicated with colors other than RGB or CMY."

It is exactly the same. The only complication is trying to show it using RGB values, where the exact value of any colors other than RGB is not a single scalar. When converted to other color encodings it becomes exactly the same. In HSV or HSL color spaces the saturation is a single scalar regardless of the hue.

"The only time that high ISO gets involved is when it leads you to overexpose one or more of the primary RGB values in the raw file."

That is one instance, but there are others. It also depends on the gamma curves used. And it depends on various types of noise, which also dilute the color purity. Obviously at high ISO where read noise is a greater portion of the data that can be a problem if not handled correctly. Photon noise can affect saturation at any brightness level.

Reply
Page 1 of 5 next> last>>
If you want to reply, then register here. Registration is free and your account is created instantly, so you can post right away.
Main Photography Discussion
UglyHedgehog.com - Forum
Copyright 2011-2024 Ugly Hedgehog, Inc.