[ROBEFFY]

Hi Normhead,
That is the creative component, or practical judgement kicking in. You make a good point.
I agree, at times there is no point to avoiding a blown out spot or pure dark area. I want pure dark in the back of a cave, or blown highlight in the reflection off a bumper.
My personal experience has had very little benefit from the histogram, but there have been tricky lighting situations where I occasionally refer to it, and will make an adjustment and take another picture.

[zekewhipper]

Don't.
I never found a use where it gave adequate results. It is a representation of the JPG preview so it is hardly a fair or accurate representation of the actual image.
Assuming you shoot in raw. I could see where it may be more useful if you shoot directly to jpeg.

No, the histogram is not based on the jpeg, but literally on what the sensor captured.

[dlacouture]

Nope, it's based on the jpg... Easy way to check: develop the same file in-camera using different white balances or exposure settings...

There is only one way IMO to use ETTR in RAW : UniWB...
Is quite easy to set up :
* create a custom white balance using a completely overexposed image (pure white), the camera will then lose its marbles and apply a 1-1-1 conversion ratio on its RGB channels (resulting in a greenish image due to the RGBG Bayer filter).
* tweak contrast and luminosity settings so the histogram gets similar to what you can have using a software like Rawnalyze.

And voilà! a near-perfect RAW histogram... Only downside : all your pics are now green, so RAW+ is out of question...

[zekewhipper]

Nope, it's based on the jpg... Easy way to check: develop the same file in-camera using different white balances or exposure settings...

There is only one way IMO to use ETTR in RAW : UniWB...
Is quite easy to set up :
* create a custom white balance using a completely overexposed image (pure white), the camera will then lose its marbles and apply a 1-1-1 conversion ratio on its RGB channels (resulting in a greenish image due to the RGBG Bayer filter).
* tweak contrast and luminosity settings so the histogram gets similar to what you can have using a software like Rawnalyze.

And voilà! a near-perfect RAW histogram... Only downside : all your pics are now green, so RAW+ is out of question...

Again, no it is not. That is why the image you see on your LCD can vary so much exposure-wise from what you actually see on your computer screen. Which then again emphasizes why your exposure should be based on the histogram and not the LCD.

[dlacouture]

And again: yes, it's based on the jpg preview...

Proof again (and I checked this on my K5!):
- take a RAW photo in a given WB
- develop this photo in-camera using the same WB (you should get the same photo)
- compare the histograms

If what you say is true, the histograms should be different, as the JPG histogram is always different from RAW, no matter what WB you use.
At least on all Pentax cameras I had laid my hands upon (the last one being the K5, so maybe the K3 differs on this), this method gives identical histograms, proof that the displayed histogram of a RAW file is in fact the JPG preview histogram.

Again, the only way to have an accurate RAW histogram is to give UniWB a try, but you lose the RAW+JPG benefits

[zekewhipper]

this method gives identical histograms, proof that the displayed histogram of a RAW file is in fact the JPG preview histogram.

Or, from another vantage point, they match because the JPEG preview histogram is the RAW histogram like I said.

[dlacouture]

You don't admit to defeat easily, huh?

Well, try this one : in manual mode, RAW, take two identical shots, only changing WB between the two (say, alternate between Shade and Tungsten)...

From your theory, histograms should be identical, no? I'll let you try and comment...

From memory, I think only Hasselblad (or was it Phase One?) does propose a true RAW histogram in their cameras...

[zekewhipper]

Why do you keep bringing up WB? We are talking about exposure histograms (not RGB histograms). They are independent of one another.


Ponder this for a second, what good does it do a photographer to be presented with a histogram on the LCD if they cannot have faith that they can use it as a decision making tool if it will give them misleading information.


Haven't you ever heard of "exposing to the right"? That only works if the histogram gives you an accurate representation of the exposure. It can only do that if it is based on the RAW data.


Furthermore, if you are expecting the histogram as shown on your camera's LCD to be an exact match to what you see on your computer screen, that is not going to happen. They are two entirely different screens with their own proportional format sizes, data presentation method defaults, and resolutions. You can't expect to overlay the two.

[BrianR]

Why do you keep bringing up WB? We are talking about exposure histograms (not RGB histograms). They are independent of one another.

WB changes the jpeg preview, which then changes the exposure histogram. The exposure histogram does not appear to be independent of the individual colour channels (on the k5iis at least).

Have you tried two identical raw shots with drastically different WB? Below is what you'll get (k5iis aimed at a solid green target). The exposure histogram changes a good deal (not as much as the RB arts of the RGB of course). If I throw these two raw files into lightroom and change their WB to something common, the histograms then match each other in lightroom.

Same raw file except different WB setting, different exposure histogram on the back of the camera. The exposure histogram is not based solely on the raw data captured.

You can try a similar experiment changing the saturation, contrast, etc. jpeg settings while still shooting in raw.

[zekewhipper]

WB changes the jpeg preview, which then changes the exposure histogram. The exposure histogram does not appear to be independent of the individual colour channels (on the k5iis at least).

Have you tried two identical raw shots with drastically different WB? Below is what you'll get (k5iis aimed at a solid green target). The exposure histogram changes a good deal (not as much as the RB arts of the RGB of course). If I throw these two raw files into lightroom and change their WB to something common, the histograms then match each other in lightroom.

Same raw file except different WB setting, different exposure histogram on the back of the camera. The exposure histogram is not based solely on the raw data captured.

You can try a similar experiment changing the saturation, contrast, etc. jpeg settings while still shooting in raw.

I see what you posted, and can agree that changing the WB can change the histogram. That is of course because you have radically changed a color channel by putting the WB dramatically off target. In essence you are trying to force the shade (lightness & darkness ergo exposure) of the color to be something it is not. The overwhelming majority of photographers don't go about shifting their WB to such extremes from real world color.


When I said independently, I was talking about in terms of photo creation. I don't know of any photographer that is thinking about the separate color channels having to be just so to get a proper exposure. Generally one tackles getting the WB colors right first, then they approach adjusting the exposure as necessary to get it correct. Mentally they treat the two things independently.


Try this experiment. Take a picture of a plain white surface. Increase your compensation so that it appears the same brightness as -you- see it. Now set your camera to Manual mode to the same settings as the last shot including WB, and take the picture again but as a JPEG. Then, using all the same settings including the WB, take the shot again but in RAW. You will notice that the histograms match.


I have never encountered any other photogs that regard WB and exposure as being related. They treat them as independent entities.


As for exposure histogram accuracy, as long as you have strived for accurate color, then whether you believe the exposure histogram is JPEG or RAW based won't really matter, because it still will lead you to draw the same conclusions about what is going on with the image exposure-wise.

[BrianR]

I see what you posted, and can agree that changing the WB can change the histogram. That is of course because you have radically changed a color channel by putting the WB dramatically off target. In essence you are trying to force the shade (lightness & darkness ergo exposure) of the color to be something it is not. The overwhelming majority of photographers don't go about shifting their WB to such extremes from real world color.

I wasn't claiming it was a reasonable real world thing. Just that it demonstrates the histogram is not based directly on the raw data but on the embedded jpeg preview, which is dependent on the various settings that affect the jpeg, WB, saturation, etc.

Try this experiment. Take a picture of a plain white surface. Increase your compensation so that it appears the same brightness as -you- see it. Now set your camera to Manual mode to the same settings as the last shot including WB, and take the picture again but as a JPEG. Then, using all the same settings including the WB, take the shot again but in RAW. You will notice that the histograms match.

Yes, of course they will.

[zekewhipper]

I wasn't claiming it was a reasonable real world thing. Just that it demonstrates the histogram is not based directly on the raw data but on the embedded jpeg preview, which is dependent on the various settings that affect the jpeg, WB, saturation, etc.

It doesn't demonstrate that definitively at all. There is nothing that says your LCD can't give you a JPEG preview image along with a RAW based histogram.

[BrianR]

It doesn't demonstrate that definitively at all. There is nothing that says your LCD can't give you a JPEG preview image along with a RAW based histogram.

The captured RAW data was the same in both the pictures and yet the exposure histogram was different. The WB setting was the only difference and you can get a similar result with the other jpeg settings of saturation, contrast, and so on. You may believe whatever you wish from the result, but it's pretty clear to me.

[zekewhipper]

The captured RAW data was the same in both the pictures and yet the exposure histogram was different. The WB setting was the only difference and you can get a similar result with the other jpeg settings of saturation, contrast, and so on. You may believe whatever you wish from the result, but it's pretty clear to me.

I'm moving on too.

[dlacouture]

Try this experiment. Take a picture of a plain white surface. Increase your compensation so that it appears the same brightness as -you- see it. Now set your camera to Manual mode to the same settings as the last shot including WB, and take the picture again but as a JPEG. Then, using all the same settings including the WB, take the shot again but in RAW. You will notice that the histograms match.

... which is perfect proof that the histogram of a RAW image is in fact taken from its JPG preview pic, I really don't see how you can be persuaded otherwise!

If the contrary was true (the JPG histogram being based on the RAW data), then it would mean that the histogram would be useless for JPG shooters (as it would be based on data from a format not even recorded)...