How Important Is The Histogram . . . Really?

[xpatUSA]

Some of us exposed optimally for very unforgiving media, transparency film, for years and professionally long before Histograms existed on cameras. When I shot national ads, shot for magazines, for annual reports, all on transparency film, I needed to nail exposure by 1/4 a stop or I didn't deserve to be paid! You absolutely do NOT need any Histogram on your camera to do this. Unfortunately, a lot of 'young folks' never learned fundamental exposure, something that's photography 101 without the lie of the Histogram as perhaps a crutch. Would having a raw Histogram be useful on camera? Sure. Must we have one? No. But don't look at a Histogram built from and based on a camera JPEG and expect it has bearing on the raw data or is necessarily to use one to optimally expose.

Young folks do not have to care what old folks had to do back in the day, when better tools exist to do the job better, like histograms. Histogram is a tool and using one does not make it a crutch or a lack of knowledge. A raw histogram would be great, the "lie" based on the camera preview is good enough if you spend a little time understanding how the in camera histogram on your specific camera correlates with the raw data.

There is no raw histogram.

Incredible!

a rgb histogram is made with rgb values. Raw is made of linear lab values. (and with a gama 1 I guess). I guess you can make one, but it would be no significant.

Of no significance? Ridiculous.

An RGB histogram is dependent upon an RGB system AND a gamut profile (sRGB/Adobe/ProPhoto, etc).

No comment other than please repair your shift key. ;-)

 

[MauriceRR]

There is no raw histogram.

Incredible!

What you see is a output rvb histogram, not an input histogram. I guess with an appropriate tool you can have histogram from a raw, just let me show one, specially from a x3f.

a rgb histogram is made with rgb values. Raw is made of linear lab values. (and with a gama 1 I guess). I guess you can make one, but it would be no significant.

Of no significance? Ridiculous.

"irrelevant" would be more exact. Sorry for my bad English.

Edit: I mean, I don't know for you, but I'm used to work with rgb histo, and a histogram on a achannel from a lab system with a gamma which is not 2.2, It's strongly hard to interprate and to use it for retouching...

An RGB histogram is dependent upon an RGB system AND a gamut profile (sRGB/Adobe/ProPhoto, etc).

No comment other than please repair your shift key. ;-)

I wanted to emphase the difference of a system (RGB/LAB/CMJN, etc) and a profile inside a system (AdobeRGB, Fogra27, etc).

 

[xpatUSA]

There is no raw histogram.

Incredible!

What you see is a output rvb histogram, not an input histogram. I guess with an appropriate tool you can have histogram from a raw, just let me show one, specially from a x3f.

This is a raw histogram from the original X3F for the red flower:



I see it very well. So, perhaps you meant that there is no raw histogram in the camera and not in SPP either.

Some applications can show a raw histogram, I think UFraw for example.

P.S. your English is much better than my schoolboy French !!

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[MauriceRR]

This is a raw histogram from the original X3F for the red flower:



I see it very well. So, perhaps you meant that there is no raw histogram in the camera and not in SPP either.

Ok, I tested raw digger, you are right, you can make one. not bad, It create input relevant histogram, not in a rgb coding system(0-255values) and the ev0 is no sense to me, but you can see the data with precision. A bit hard to interprate, but it's ok with the option non linear abciss for those who are use to see rvb histo.

Meanwhile, for retouching an image, you need to know what's happening in the conversion into a rgb profile. I can't really do something with rawdigger for that, it lacks of tools, but it's something which would be a welcome intregration in the adobe suite, with the output histogram

 

[xpatUSA]

This is a raw histogram from the original X3F for the red flower:



I see it very well. So, perhaps you meant that there is no raw histogram in the camera and not in SPP either.

Ok, I tested raw digger, you are right, you can make one. not bad, It create input relevant histogram, not in a rgb coding system(0-255values) and the ev0 [makes] no sense to me

I usually set the EV 0 at the level which saturate the sensor. For Scott's SD14, that is approximately 8000. Therefore it possible to discuss exposure headroom for example.

SD14 raw data is not as easy to examine, unlike SD1 Merrill raw data. With the SD1M, the data has a hard limit at about 4080 - just short of the theoretical 4095. With the SD14, the metadata values for saturation are about 6000 but gross over-exposure can give levels over 9000! This is called a 'soft' limit or it can be called non-linear in higher levels, like 'S' curve for film?

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[MauriceRR]

It makes me think another thing. There is obviously a EV compression which is done in a raw software, (or camera) from the input to the output. How many EV you can put in a let's say srgb profile ? (without tweaking contrast/whites and darks).

 

[digidog]

This is a raw histogram from the original X3F for the red flower:



I see it very well. So, perhaps you meant that there is no raw histogram in the camera and not in SPP either.

Ok, I tested raw digger, you are right, you can make one. not bad, It create input relevant histogram, not in a rgb coding system(0-255values) and the ev0 is no sense to me, but you can see the data with precision. A bit hard to interprate, but it's ok with the option non linear abciss for those who are use to see rvb histo.

Meanwhile, for retouching an image, you need to know what's happening in the conversion into a rgb profile.

That's after proper exposure for the raw data, IOW, not pertinent but rather important for dealing with color clipping when picking an encoding color space (sRGB vs. ProPhoto RGB).

I can't really do something with rawdigger for that, it lacks of tools, but it's something which would be a welcome intregration in the adobe suite, with the output histogram

--
Andrew Rodney
Author: Color Management for Photographers
The Digital Dog

Home

 

[Scottelly]

There is no raw histogram.

Incredible!

What you see is a output rvb histogram, not an input histogram. I guess with an appropriate tool you can have histogram from a raw, just let me show one, specially from a x3f.

This is a raw histogram from the original X3F for the red flower:

Thank you for posting this histogram Ted. It looks to me like there is lots of green and blue! I wonder why green is showing as 0 almost all over the flower.

I see it very well. So, perhaps you meant that there is no raw histogram in the camera and not in SPP either.

Some applications can show a raw histogram, I think UFraw for example.

P.S. your English is much better than my schoolboy French !!

Je parle un peut de Francais aussie!

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

--
Scott Barton Kennelly

http://www.bigprintphotos.com

 

[Scottelly]

It makes me think another thing. There is obviously a EV compression which is done in a raw software, (or camera) from the input to the output. How many EV you can put in a let's say srgb profile ? (without tweaking contrast/whites and darks).

I get the feeling as though there could be a lot more manual control over things offered in SPP. Maybe they could create an "advanced" mode, which could be turned on, and then we could have manual control over curves and all sorts of other stuff. SPP could go far!



BTW, I'd really like to see cropping added, since the files can be cropped to different dimensions in camera, after-all. It would be nice to add arbitrary rotation at the same time (not just 90 degree rotation), for straightening up images that have tilted horizons and such.

 

[MauriceRR]

That's after proper exposure for the raw data, IOW, not pertinent but rather important for dealing with color clipping when picking an encoding color space (sRGB vs. ProPhoto RGB).

I wish i could, but i can't say from raw digger's histogramm that i need prophoto space or I can go with adobe or srgb space.

Or I missed something in the programm ?

I just can say that there is no clipped channel in the raw. Which is not truely interesting because It can produce positive false : a clipped channel only won't make a bad exposure : one channel can produce texture, the same way there is still textures in the srgb files from the red flower, meanwhile green and blue are clipped.

 

[digidog]

That's after proper exposure for the raw data, IOW, not pertinent but rather important for dealing with color clipping when picking an encoding color space (sRGB vs. ProPhoto RGB).

I wish i could, but i can't say from raw digger's histogramm that i need prophoto space or I can go with adobe or srgb space.

That isn't it's role or intent. The encoding color space (sRGB, ProPhoto) can be evaluated in any raw converter that shows color clipping as I illustrated in my ACR video. And this has to happen AFTER the raw data is rendered in the native color space of the converter.

Or I missed something in the programm ?

Again, you're looking at raw data. There IS NO color gamut yet. Digital cameras and scanners do not have a colorimetric color gamut. So this application isn't showing you anything to do with color gamut and everything to do with exposure of the raw data.

I just can say that there is no clipped channel in the raw.

You can be told if there is clipping yes, due to over exposure. You can't be told if there is color clipping; there's no color, per se, yet.

Which is not truely interesting because It can produce positive false : a clipped channel only won't make a bad exposure : one channel can produce texture, the same way there is still textures in the srgb files from the red flower, meanwhile green and blue are clipped.

You need to define (and understand) what kind of clipping is under discussion. One can clip colors due to too small a color gamut for encoding and again, I illustrated that in my video! You can see color clipping from raw to sRGB in ACR but NOT when one switches to encoding into ProPhoto RGB.

--
Andrew Rodney
Author: Color Management for Photographers
The Digital Dog
http://www.digitaldog.net

 

[MauriceRR]

Thank you for posting this histogram Ted. It looks to me like there is lots of green and blue! I wonder why green is showing as 0 almost all over the flower.

Because of the srgb files you did probably. The raw space is a lot bigger than you srgb files with the clipped channel.

Moreover it's an illustration of what I tryed to say, this histogramm is not really relevant, because the traduction in a // RVB system + space + 2.2gamma space relevant) will show a fractionof this histogram + other proportions.

It would be a nice challenge to have the same fil with a bayer sensor, to see if would see also green infos in the raw. With a bit of imagination I would say hardly.

I guess this test would be more challenging for a bayer sensor, but maybe not with a so big importance, cause we are very far of what we can print, or even display in adobe rgb. So the unique channel clipping is... not really meaningfull for me. More challenging is to soft proof something like this flower.

Je parle un peut de Francais aussie!

pas mal !

 

[MauriceRR]

I just can say that there is no clipped channel in the raw.

You can be told if there is clipping yes, due to over exposure. You can't be told if there is color clipping; there's no color, per se, yet.

Yeah, I don't know Ho to call it, I called it channel. But I still see 3 colored histogramms

Which is not truely interesting because It can produce positive false : a clipped channel only won't make a bad exposure : one channel can produce texture, the same way there is still textures in the srgb files from the red flower, meanwhile green and blue are clipped.

You need to define (and understand) what kind of clipping is under discussion. One can clip colors due to too small a color gamut for encoding and again, I illustrated that in my video! You can see color clipping from raw to sRGB in ACR but NOT when one switches to encoding into ProPhoto RGB.

Ok, So the raw can't clipp just for one color ? Let's imagine a color far away even from prophoto gamut ? The raw clipp only for luminosity ? Im' not sure to really understand the system under the raw file. Nether if we are only talking about foveon or for every raw files including bayers.

 

[digidog]

I just can say that there is no clipped channel in the raw.

You can be told if there is clipping yes, due to over exposure. You can't be told if there is color clipping; there's no color, per se, yet.

Yeah, I don't know Ho to call it, I called it channel. But I still see 3 colored histogramms

OF raw data. It's grayscale at this state to simplify the conversation.

Which is not truely interesting because It can produce positive false : a clipped channel only won't make a bad exposure : one channel can produce texture, the same way there is still textures in the srgb files from the red flower, meanwhile green and blue are clipped.

You need to define (and understand) what kind of clipping is under discussion. One can clip colors due to too small a color gamut for encoding and again, I illustrated that in my video! You can see color clipping from raw to sRGB in ACR but NOT when one switches to encoding into ProPhoto RGB.

Ok, So the raw can't clipp just for one color ?

This product isn't showing us the options for an encoding color space so color clipping based on encoding is moot. It hasn't happened (yet).

Let's imagine a color far away even from prophoto gamut ?

It would not be a color. If you can't see it, it's not a color.

The raw clipp only for luminosity ?

Yes it can. Or not if 'ideally exposed'. That's what RawDigger shows us.

Im' not sure to really understand the system under the raw file. Nether if we are only talking about foveon or for every raw files including bayers.

Raw is raw; it's not encoded into any fixed color space so we can't talk about a color space at this stage. We can talk about exposure's effect on the raw data!

 

[MauriceRR]

It would not be a color. If you can't see it, it's not a color.

Really ?

I like it already....

The raw clipp only for luminosity ?

Yes it can. Or not if 'ideally exposed'. That's what RawDigger shows us.

ok

Im' not sure to really understand the system under the raw file. Nether if we are only talking about foveon or for every raw files including bayers.

Raw is raw; it's not encoded into any fixed color space so we can't talk about a color space at this stage. We can talk about exposure's effect on the raw data!

ok.

I found your videos, I'm going to take a look.

 

[digidog]

It would not be a color. If you can't see it, it's not a color.

Really ?

Indeed!

Color, is a perceptual property. So if you can't see it it's not a color. Color is not a particular wavelength of light. It is a cognitive perception, the excitation of photoreceptors followed by retinal processing and ending in the our visual cortex, within our brains. As such, colors are defined based on perceptual experiments.

Don't take my word for this alone; here are others who understand the color science:


Fairchild's "Color Appearance Models". Page 1! Like beauty, color is in the eye of the beholder. For as long as human scientific inquiry has been recorded, the nature of color perception has been a topic of great interest. Despite tremendous evolution of technology,fundamental issues of color perception remain unanswered. Many scientific attempts to explain color rely purely on the physical nature of light and objects. However, without the human observer, there is no color.

Further on the same page:

It is common to say that certain wavelengths of light, or certain objects are a give color. This is an attempt to relegate color to the purely physical domain. It is more correct to state those stimuli are perceived to be a certain color when viewed under specific conditions.

Page 1 paragraph 2 of Digital Color Management by Giorgianni and Madden:

But color itself is a perception and perceptions only exist in the mind.

Page 11 of The GATF Practical guide to color management:

Although extensive research has been conducted, we still not completely understand what happens in the brain when we "see" color. The visual sensation known as color occurs when light excites photoreceptors in the eye called cone cells.

Page 75 of Understanding color management by Sharma:

Color is an impression that we form in our brains.

I like it already....

The raw clipp only for luminosity ?

Yes it can. Or not if 'ideally exposed'. That's what RawDigger shows us.

ok

Im' not sure to really understand the system under the raw file. Nether if we are only talking about foveon or for every raw files including bayers.

Raw is raw; it's not encoded into any fixed color space so we can't talk about a color space at this stage. We can talk about exposure's effect on the raw data!

ok.

I found your videos, I'm going to take a look.

 

[MauriceRR]

Not sure it's truely relevant, but... I put that here, in case of.

http://www.brucelindbloom.com/index.html?LabGamutDisplayHelp.html

EDIT : Even if it's not in the lab, it does'nt mean it's not in the raw ?

So I reply to myself : like debayerised sensors ? uv photography ? etc ?

 

[xpatUSA]

I just can say that there is no clipped channel in the raw.

You can be told if there is clipping yes, due to over exposure. You can't be told if there is color clipping; there's no color, per se, yet.

And may I add that if there are no clipped raw channels - and even a good bit of headroom - the Foveon conversion matrices have pretty big coefficients including the off-diagonal ones. And I'm talking camera-to-XYZ - an even bigger space than ProPhoto as of course you know.

Sensor Inputs refers to bottom, middle and top layer signals.

An early matrix for the SD9 - later ones vary.

For example, if I put equal layer exposures, with 1 EV or so headroom, into the Sigma SD14 cam-XYZ matrix, I would naively expect to get something near a white point out of the matrix. But no, it gives XYZ = 0.55, 0.39, 0.91 and that transforms to an x,y chromaticity of 0.3, 0.21- placing the white point nowhere near neutral, more a light purple I reckon.

Main point, since I digressed, is that lack of clipping in the raw is no guarantee that the RGB will not be clipped.

--
"What we've got hyah is Failyah to Communicate": 'Cool Hand Luke' 1967.
Ted

 

[digidog]

Not sure it's truely relevant, but... I put that here, in case of.

http://www.brucelindbloom.com/index.html?LabGamutDisplayHelp.html

It's not really relevant. ALL color spaces can be mapped in reference to Lab gamut. And Lab is based on the perception of the Human Observer.

EDIT : Even if it's not in the lab, it does'nt mean it's not in the raw ?

It means it's not a color!

Again, raw files have no colorimetric color gamut.

Loading…

www.color-image.com

And even color spaces that have a gamut can define numbers that are not colors (we can't see them):

http://digitaldog.net/files/ColorNumbersColorGamut.pdf

R0/G255/B0 in ProPhoto RGB is NOT a color!

 

[digidog]

I just can say that there is no clipped channel in the raw.

You can be told if there is clipping yes, due to over exposure. You can't be told if there is color clipping; there's no color, per se, yet.

And may I add that if there are no clipped raw channels - and even a good bit of headroom - the Foveon conversion matrices have pretty big coefficients including the off-diagonal ones. And I'm talking camera-to-XYZ - an even bigger space than ProPhoto as of course you know.

Sensor Inputs refers to bottom, middle and top layer signals.

An early matrix for the SD9 - later ones vary.

For example, if I put equal layer exposures, with 1 EV or so headroom, into the Sigma SD14 cam-XYZ matrix, I would naively expect to get something near a white point out of the matrix. But no, it gives XYZ = 0.55, 0.39, 0.91 and that transforms to an x,y chromaticity of 0.3, 0.21- placing the white point nowhere near neutral, more a light purple I reckon.

Main point, since I digressed, is that lack of clipping in the raw is no guarantee that the RGB will not be clipped.

Further, the rendering settings in a raw converter play a huge role here! Take a raw capture of a scene that itself has a very wide color gamut. Yank down Saturation (or vibrance), it can easily be encoded into sRGB without any color clipping.

--
Andrew Rodney
Author: Color Management for Photographers
The Digital Dog
http://www.digitaldog.net