2007-02-11 02:01:26
Is there a universal ideal histogram, say a
bell-shaped distribution?
bell-shaped distribution?
#1
Is there a universal ideal histogram, say a
bell-shaped distribution?
In article <eqltsi$6uv$>,
"Peter Jason" wrote:
Is there a universal ideal histogram, say a
bell-shaped distribution?
No. The histogram will always depend on the specifics of the image. An image with one strong predominant tone will have a histogram that reflects this; the histogram of a high-contrast image will be different from the histogram of a low-contrast image; and so on.
tacit wrote:
In article <eqltsi$6uv$>,
"Peter Jason" wrote:
Is there a universal ideal histogram, say a
bell-shaped distribution?
No. The histogram will always depend on the specifics of the image. An image with one strong predominant tone will have a histogram that reflects this; the histogram of a high-contrast image will be different from the histogram of a low-contrast image; and so on.
Correct. Many people don't realise this, but the histogram *is* your image. It's just another way of presenting the data. Shoot a snowy landscape, and your histogram should be and will be primarily on the right, because most pixels will be white. Shoot a starry night, and 99% of the histogram is on the far left (the dark sky) and 1% (the stars) is on the far right.
On 2007-02-11 05:57:47 -0400, (Johan W. Elzenga) said:
tacit wrote:
In article <eqltsi$6uv$>,
"Peter Jason" wrote:
Is there a universal ideal histogram, say a
bell-shaped distribution?
No. The histogram will always depend on the specifics of the image. An image with one strong predominant tone will have a histogram that reflects this; the histogram of a high-contrast image will be different from the histogram of a low-contrast image; and so on.
Correct. Many people don't realise this, but the histogram *is* your image. It's just another way of presenting the data. Shoot a snowy landscape, and your histogram should be and will be primarily on the right, because most pixels will be white. Shoot a starry night, and 99% of the histogram is on the far left (the dark sky) and 1% (the stars) is on the far right.
The only thing that you really shoot for in the histogram is to stay away from "combing".
Correct. Many people don't realise this, but the histogram *is* your image. It's just another way of presenting the data. Shoot a snowy landscape, and your histogram should be and will be primarily on the right, because most pixels will be white. Shoot a starry night, and 99% of the histogram is on the far left (the dark sky) and 1% (the stars) is on the far right.
The only thing that you really shoot for in the histogram is to stay away from "combing".
It's a bit different for scanning ...
Your histogram here will also show you if your scanner is presetting the correct range for the document. IF it's filled to either extreme, then the scanner's range setting is too narrow and some information will be lost.
Sir F. A. Rien wrote:
Correct. Many people don't realise this, but the histogram *is* your image. It's just another way of presenting the data. Shoot a snowy landscape, and your histogram should be and will be primarily on the right, because most pixels will be white. Shoot a starry night, and 99% of the histogram is on the far left (the dark sky) and 1% (the stars) is on the far right.
The only thing that you really shoot for in the histogram is to stay away from "combing".
It's a bit different for scanning ...
Your histogram here will also show you if your scanner is presetting the correct range for the document. IF it's filled to either extreme, then the scanner's range setting is too narrow and some information will be lost.
Not if you scan that starry night picture...
It's a bit different for scanning ...
Your histogram here will also show you if your scanner is presetting the correct range for the document. IF it's filled to either extreme, then the scanner's range setting is too narrow and some information will be lost.
Not if you scan that starry night picture...
Yes, -=IF=- the scanner's "automatic" settings clip off the blacks so that you lose the very subtle shadow detail.
Sir F. A. Rien wrote:....
It's a bit different for scanning ...
Your histogram here will also show you if your scanner is presetting the
correct range for the document. IF it's filled to either extreme, then the
scanner's range setting is too narrow and some information will be lost.
Not if you scan that starry night picture...
Yes, -=IF=- the scanner's "automatic" settings clip off the blacks so that
you lose the very subtle shadow detail.
My point is that not every photo has this subtle shadow detail. A starry night picture can have a lot of truely black pixels, even if the scanner doesn't clip anything. And the stars will have (and should have) truely white pixels, again even without any clipping at all. You can set the scanner software so that the night sky won't become completely black or the stars won't become completely white, but that won't add any detail to that sky or those stars. If there is no detail in the film, the scanner can't create it either.
"Johan W. Elzenga" wrote in message
Sir F. A. Rien wrote:...
It's a bit different for scanning ...
Your histogram here will also show you if your scanner is presetting the
correct range for the document. IF it's filled to either extreme, then the
scanner's range setting is too narrow and some information will be lost.
Not if you scan that starry night picture...
Yes, -=IF=- the scanner's "automatic" settings clip off the blacks so that
you lose the very subtle shadow detail.
My point is that not every photo has this subtle shadow detail. A starry night picture can have a lot of truely black pixels, even if the scanner doesn't clip anything. And the stars will have (and should have) truely white pixels, again even without any clipping at all. You can set the scanner software so that the night sky won't become completely black or the stars won't become completely white, but that won't add any detail to that sky or those stars. If there is no detail in the film, the scanner can't create it either.
Good point and well said. I would add that there is no detail in a black night sky, or in a star. It is the photographer, or the person doing the final color adjustment, who determines what should be pure black, or pure white, not the histogram.
Make sure you'r scanning at 16 bit not 8 if you want more detail. even if you down to 8-bit later the 16-bit will give you much more info to color correct.
Is there a universal ideal histogram, say a
bell-shaped distribution?
In article
<eqltsi$6uv$>,
"Peter Jason" wrote:
Is there a universal ideal histogram, say
a
bell-shaped distribution?
No. The histogram will always depend on the
specifics of the image. An
image with one strong predominant tone will
have a histogram that
reflects this; the histogram of a
high-contrast image will be different
from the histogram of a low-contrast image;
and so on.
--
Photography, kink, polyamory, shareware,
and more: all at
http://www.xeromag.com/franklin.html
"Papa Joe" <Sorry> wrote in message news:2007021216321750073- ...
Make sure you'r scanning at 16 bit not 8 if you want more detail. even if you down to 8-bit later the 16-bit will give you much more info to color correct.
Not necessarily. Though many people prefer to scan in 16 bit, this is not generally necessary. There is a lot of headroom in an 8 bit scan for subsequent edits. Yes, there will be gaps in the histogram, but these gaps are not an indicator of lack of image quality. People buy images, not histograms.
Thanks, I am photographing architecture,
usually in the very early morning as the sun
is rising. The sky tends to be much brighter
than the buildings. Therefore should I aim
for a histogram with less on the right side
(darker) and maintain this histogram
arrangement for all the other pictures?
"Johan W. Elzenga" wrote in message
Sir F. A. Rien wrote:
It's a bit different for scanning ...
Your histogram here will also show you if your scanner is presetting the
correct range for the document. IF it's filled to either extreme, then the
scanner's range setting is too narrow and some information will be lost.
Not if you scan that starry night picture...
Yes, -=IF=- the scanner's "automatic" settings clip off the blacks so that
you lose the very subtle shadow detail.
My point is that not every photo has this subtle shadow detail. A starry night picture can have a lot of truely black pixels, even if the scanner doesn't clip anything. And the stars will have (and should have) truely white pixels, again even without any clipping at all. You can set the scanner software so that the night sky won't become completely black or the stars won't become completely white, but that won't add any detail to that sky or those stars. If there is no detail in the film, the scanner can't create it either.
Good point and well said. I would add that there is no detail in a black night sky, or in a star. It is the photographer, or the person doing the final color adjustment, who determines what should be pure black, or pure white, not the histogram.
Hint: The histogram is only a TOOL to allow the person to judge the scan. Once clipped by accepting an 'auto' setting, the information is gone, gone,
gone !!!
"Sir F. A. Rien" wrote in message
...
Hint: The histogram is only a TOOL to allow the person to judge the scan. Once clipped by accepting an 'auto' setting, the information is gone, gone,
gone !!!
Ca ne fait rien, in many cases.
It is important to clip, or at least compress, unimportant shadow and highlight detail. This adds contrast and tonal range to the more important parts of the image.
The histogram is of no help in determining which detail is important. It is too egalitarian. By treating all data equally, the histogram tool fosters the impression that all highlight and shadow detail is equally important. It is not.
The world would be a better place, in terms of image quality, if the histogram had never been discovered. It is far better to rely on the image itself, and using the info tool to judge highlight and shadow values.
Mike Russell
www.curvemeister.com/forum/
"Papa Joe" <Sorry> wrote in message news:2007021216321750073- ...
Make sure you'r scanning at 16 bit not 8 if you want more detail. even if you down to 8-bit later the 16-bit will give you much more info to color correct.
Not necessarily. Though many people prefer to scan in 16 bit, this is not generally necessary. There is a lot of headroom in an 8 bit scan for subsequent edits. Yes, there will be gaps in the histogram, but these gaps are not an indicator of lack of image quality. People buy images, not histograms.
--
Mike Russell
www.curvemeister.com/forum/
Hi Mike,
The 16 bit image has extra color information that really help in the smoothen of tones and that is image quality. The 8 bit image will suffer posterization of tones on the hot spot area of a subject upon editing.
One good example of hot spot area is the highlight on the model's forehead. The 16 bit image is smooth in this case. Off course 8 bit color depth is sufficient if the diff in tones luminosity within a specific area is not great.
Peter Jason wrote:
Thanks, I am photographing architecture,
usually in the very early morning as the sun
is rising. The sky tends to be much brighter
than the buildings. Therefore should I aim
for a histogram with less on the right side
(darker) and maintain this histogram
arrangement for all the other pictures?
No, you should aim for a good image. Don't let the histogram tell you what a good image is, let your eyes tell you. The histogram is a good aid to tell you what is going on with your image and how you could possibly improve it. But in the end, the histogram means nothing if your eyes tell you the image is good.
--
Johan W. Elzenga johan<<at>>johanfoto.nl Editor / Photographer http://www.johanfoto.com
and these new shooters who use all auto settings all the time with a Canon Rebel and 50mm lens and call themselves a pro....
AFAIK there is no image referred to in this thread. If you have a link, I'd like to take a look at what you are talking about.
"mirafiori" wrote in message
Hi Mike,
The 16 bit image has extra color information that really help in the smoothen of tones and that is image quality. The 8 bit image will suffer posterization of tones on the hot spot area of a subject upon editing.
This is the standard argument. While I have no argument with people who prefer, for their own reasons, to do everything uniformly in 16 bit, your statement about posterization of 8 bit images is simply not backed up by actual images.
One good example of hot spot area is the highlight on the model's forehead. The 16 bit image is smooth in this case. Off course 8 bit color depth is sufficient if the diff in tones luminosity within a specific area is not great.
AFAIK there is no image referred to in this thread. If you have a link, I'd like to take a look at what you are talking about.
--
Mike Russell
www.curvemeister.com/forum/
AFAIK there is no image referred to in this thread. If you have a link, I'd like to take a look at what you are talking about.
http://www.cambridgeincolour.com/tutorials/posterization.htm
OK as we are only able to see rather poorly in the camera viewer and the laptop in the sunny areas where we are working
We often use the histogram as a check for the white areas like a shirt or a forehead where the light hits
to make sure the highlights are not burnt up
it is easier to correct an image lighter in PS but if the highlights are not recorded on the CCd they are gone
is that a good reason to use histogram??
of course my photographer is old fashioned and still depends upon the handheld light meter readings as the be all end all!!
this week he compared it to the all auto settings and it wasn't all that different in MOST light situations.
He apprenticed under a man who insisted that if you get in a certain situation (batteries died, meter reading is off,equip failure etc) you better know it in your head.