exposure histograms

[hornetzoo]

How much faith do you put in exposure histograms? As far as I can tell, they only let you know if a picture is overexposed or underexposed, and they are only a rough estimate. But maybe if I knew how to read them better, I could tell more. any comments?

[gasrocks]

Oh gosh - you'll find people who love and use Histograms and those who don't. I think it mostly comes into play when you are on the run, wondering if you exposed the last pix correctly, and it is too light out to really see the image in the LCD. Then some will look at the histogram for added info. I don't use them myself. But, I have many students who cannot tell correct exposure by looking at a pix (duh) but see it is wrong when the histogram looks "funny."

[ssim]

When I start shooting something I always look at my histogram the first few shots to make sure that I am not blowing the highlights in general. This is about the only time I actually look at the lcd display on the camera. I know what I want the histogram to look like and once there by tweaking the exposure levels on the camera I rarely go back and look at it.

I certainly can't look at just the image on the lcd preview and tell whether or not it is good or not. I've looked at some that I thought were good and then you get the images downloaded and they look like crap.

[Moments]

The histogram is the best way to evaluate the exposure of any digital image file. Sure if you really know your display you can tell if an image is over or under, but the histogram is the best way. You need to learn how to read one.

At a reception hall that I was at one day, the photographer from the next room and I were talking in the hall about digital since he was realativly new to digital. He started to say how his lab is telling him he is usually underexposed on most of his images and he did not understand why. "it looks great on my cameras display" he said. I asked him what the historgram shows. He asked "is that the display with all the peaks and lines?" Once I showed him how to get to see the histogram, I showed him that the shot he did of the bride in her white dress was underexposed. The brightest part of the histogram was just past the midpoint. He had made his display brighter so the images looked right.

Again, someone who decided to shoot a digital wedding, without knowing how to really use a digital camera, at the expense of the Bride and Groom.

[tim]

The histogram is an essential tool, once you understand how to use it. Without it you're relying completely on the camera's metering, which is far from perfect because it doesn't know what you're trying to achieve.

[Todd Jacobsen]

How much faith do you put in exposure histograms? As far as I can tell, they only let you know if a picture is overexposed or underexposed, and they are only a rough estimate. But maybe if I knew how to read them better, I could tell more. any comments?

Histogram shows exposure and clipping - based on the white balance selected (whether that be AWB, or CWB).

It doesn't do well for tint nor provides insight into whether the shot was in focus.

But your ability to discern tint and focus utilizing the LCD is very limited as well.

[Titus213]

Check this out:

http://www.luminous-landscape.com/tutorials/understanding-series/understanding-histograms.shtml

[J Rabin]

Hornetzoo. It took me about a year to understand histograms well. The tutorials on Steve Hoffman's web site and Luminous Landscape are good, though "exposing to the right," should not be a religion, especially for "people" photography. Study those two. If you do portrait work, using a 10 second PhotoShop "FACE MASK" histogram as recomended from ShootSmarter.com is an amazing post shoot tool!
Anyway, in Canon dSLRs, the histogram is really a "luminance" or brightness histogram derived from the green color channel only (since there are twice as many green pixels as others, this is a good proxy for the whole image). So, it does not tell you perfectly about exposure of red and blue channels. But, darn it's the best, funnest tool I've ever used for high contrast outdoor work. It truly is the "21st Century Lightmeter" after you spend time with it. Shooting RAW, there is somewhere between 1-2 or even more stops of recoverable information even if the histogram shows clipping. A WHOPPING amount of photographic information. Shooting JPG, all this is lost. I went RAW two months after going digital. Spot metering and light meters are still essential for many people. But, for "on-the-go" casual photography, the histogram really helps get it right, get it better, and teaches you how to read scenes. Hope that helps. Jack

[tim]

Expose to the right, but don't blow the highlights of what's important in the picture. With RAW you can recover shadows more easily than you can fix overexposed areas.

If you do portrait work, using a 10 second PhotoShop "FACE MASK" histogram as recomended from ShootSmarter.com is an amazing post shoot tool!

Do you have a link to this?

[mrclark321]

Thanks Titus213...another great learning tool!!
Another one added to my favorites...
Dan

Check this out:

http://www.luminous-landscape.com/tutorials/understanding-series/understanding-histograms.shtml

[J Rabin]

Expose to the right, but don't blow the highlights of what's important in the picture. With RAW you can recover shadows more easily than you can fix overexposed areas. Do you have a link to this?

Tim. Here is the link. It is under an instructor named Will Crockett.
http://shootsmarter.com/index.html
They pulled it from general view and now ask users to sign in, like this forum. No biggie. The technique is so in-your-face easy, and works so well in workflow, for selecting best among bracketed shots, you'll wonder why everyone doesn't teach it!
Enjoy, Jack

[sdommin]

Some people think that histograms are the best thing since sliced bread. I'm not one of them.

I don't think I've ever used a histogram to determine exposure. There's a much better method - use the LCD (either after the shot or before it, if your camera allows)! Of course its not perfect, but your "keeper" percentage will be much higher if you calibrate your eyeball to the LCD and don't worry about the histogram.

[DavidEB]

A matter of personal taste and style, as mentioned above. Some factors that have moved me to rely on the histogram are --- 1) It's the only tool I've ever used that can help me avoid blown highlights 2) The image in the review screen looks different in bright sun, cloudy days, indoors, etc... the histogram is more robust. 3) In high-contrast situations, after I've pushed the historgram to the right without blowing highights, I can judge whether there's too much lost in deep shadow and adapt to that (such as fill flash)

[robertwgross]

I don't think I've ever used a histogram to determine exposure. There's a much better method - use the LCD (either after the shot or before it, if your camera allows)! Of course its not perfect, but your "keeper" percentage will be much higher if you calibrate your eyeball to the LCD and don't worry about the histogram.

This will give you problems eventually. The LCD illumination might be set high or low, so it might be displaying the image as light or dark when the image is stored only one way.

On a dark evening, I will set my LCD to mid-bright. On a sunny day, I have it set to maximum. My histograms do not vary.

---Bob Gross---

[Qurlyjoe]

One thing to keep in mind is that the in-camera histogram is taken from the jpg thumbnail of the shot, and may or may not accurately reflect what you get with the raw data, as far as clipping at high or low end goes.

[hornetzoo]

I was looking at the "DRI" plugin. http://www.fredmiranda.com/DRI/index.html

Has anyone tried it? Please private message me if you have the plugin and would be willing to let me test it.

[J Rabin]

I was looking at the "DRI" plugin. http://www.fredmiranda.com/DRI/index.html
Has anyone tried it? Please private message me if you have the plugin and would be willing to let me test it.
I use it. What is a private message?

[hornetzoo]

J Rabin: could you please email me at hornetzoo@hotmail.com?

[KennyG]

I use it. What is a private message?

Usually someone wanting a copy of something you paid for. I forget the word, but I picture a wooden leg and parrot.

[bauerman]

I would agree that you should rely on the histogram MUCH more than the LCD display. I noticed right away after obtaining my 300D that the LCD was a big liar. I had the brightness of my LCD turned down quite a bit and pictures would look perfect on the screen - once downloaded - I was consistently disappointed with the exposures - alot of underexposures. Checking the histogram of each shot is the only real fix for this - if my histogram is slid to the left - I know that I need to "brighten" that shot up a bit and will dial in some exposure compensation.

[ohenry]

It's really easy to see who has learned photography on a film SLR and who never learned how to expose film beyond relying on the camera's automatic features. The histogram is a great tool to have, but it's NOT the best method to get proper exposure. At best, it's a backdoor method to get a decent picture...I just hope it wasn't a once in a lifetime shot that you missed because you didn't understand how to get the right exposure BEFORE the shutter is tripped.

[tim]

It's really easy to see who has learned photography on a film SLR and who never learned how to expose film beyond relying on the camera's automatic features. The histogram is a great tool to have, but it's NOT the best method to get proper exposure. At best, it's a backdoor method to get a decent picture...I just hope it wasn't a once in a lifetime shot that you missed because you didn't understand how to get the right exposure BEFORE the shutter is tripped.

The histogram is a learning tool, and a tool to check you're photos done right. After a bit of time using it people should learn how to expose a photo properly without having to check the histogram each time.

[slin100]

Anyway, in Canon dSLRs, the histogram is really a "luminance" or brightness histogram derived from the green color channel only (since there are twice as many green pixels as others, this is a good proxy for the whole image). So, it does not tell you perfectly about exposure of red and blue channels. But, darn it's the best, funnest tool I've ever used for high contrast outdoor work. It truly is the "21st Century Lightmeter" after you spend time with it. Shooting RAW, there is somewhere between 1-2 or even more stops of recoverable information even if the histogram shows clipping. A WHOPPING amount of photographic information. Shooting JPG, all this is lost. I went RAW two months after going digital. Spot metering and light meters are still essential for many people. But, for "on-the-go" casual photography, the histogram really helps get it right, get it better, and teaches you how to read scenes. Hope that helps. Jack

This is good stuff, although there are a couple of corrections. The 1-Series DSLRs can display a RGB histogram instead of luminance. Luminance is not derived only from the green channel. Green contributes about 60% to luminance, red about 30% and blue 10%.

I find the histogram very useful although the fact that it doesn't provide a histogram of the RAW data is a serious drawback. I wish Canon would address this.

[robertwgross]

Once you begin to learn histograms, you can play an interesting training exercise.

Shoot a scene, and then view it on the computer. Take a guess at what the histogram should look like, and then pencil out your guess of the shape. Then look at the actual histogram to see how close you were. It kind of trains you on how to look at a scene for lightness and darkness. Most of us look at a scene for the subject and the photographic composition.

---Bob Gross---

[PacAce]

Anyway, in Canon dSLRs, the histogram is really a "luminance" or brightness histogram derived from the green color channel only (since there are twice as many green pixels as others, this is a good proxy for the whole image). So, it does not tell you perfectly about exposure of red and blue channels.
Now this is the 2nd time I've read something like this. The first time was a few months ago at Passion for Pixels. I sure would like to know where you guys are picking up this information from because I have a Canon 1DmkII which is a "Canon dSLR", and not only does it show the histogram in luminance mode but also in RGB mode which displays the histogram in all three colors.

[Edit: Well, I'll be a son-of-a-gun. I just copied an old CRW file to a CF card and displayed the histogram for it on my 10D and sure enough, it's only displaying the histogram for the green channel. The same image in Photoshop, displayed in RGB mode has three separate spikes, one for each of the colors. On the 10D, the only spike I see is the green spike. So, it looks like I stand corrected...sort of because the green-channel only doesn't apply for the 1DmkII.

Anyway, I'd still like to know where the original source of this infomation is. :)

[J Rabin]

PacAce. Can't remember where I got the green channel from. I had too many images recoverable in PS that showed clipped on my 10D to doubt it. I think some Chuck Westfall interview post. The other day I shot with my colleague's 1dMkII. Not only not the same histogram... Forget the damn histogram, it's not the same camera. No crippled DEP mode, a real VF, averaging spot meter (even my old Oly OM-4 had this), vertical release, sensitive release, ISO in 1/3 stops. Still no ISO in VF. My 20D takes equal or better photos in MOST situations in MY hands than his casual approach. He bought it and never read the manual... I now see a person needs a MkII for serious setups (I do workplace photography) and a 20D for casual out-and-about!

[Edit: Well, I'll be a son-of-a-gun. I just copied an old CRW file to a CF card and displayed the histogram for it on my 10D and sure enough, it's only displaying the histogram for the green channel. The same image in Photoshop, displayed in RGB mode has three separate spikes, one for each of the colors. On the 10D, the only spike I see is the green spike. So, it looks like I stand corrected...sort of because the green-channel only doesn't apply for the 1DmkII.
Anyway, I'd still like to know where the original source of this infomation is. :)[/QUOTE]

[slin100]

[Edit: Well, I'll be a son-of-a-gun. I just copied an old CRW file to a CF card and displayed the histogram for it on my 10D and sure enough, it's only displaying the histogram for the green channel. The same image in Photoshop, displayed in RGB mode has three separate spikes, one for each of the colors. On the 10D, the only spike I see is the green spike. So, it looks like I stand corrected...sort of because the green-channel only doesn't apply for the 1DmkII.

Leo, the 10D histogram is luminance, not the green channel. I confirmed this by displaying the histogram on my 10D and by looking at the luminosity histogram on Photoshop CS. The green channel and luminance histogram often look the same, but I found one RAW file where they're very different.

Anyway, I'd still like to know where the original source of this infomation is. :)
I read about it on some other forum.

[Titus213]

It's really easy to see who has learned photography on a film SLR and who never learned how to expose film beyond relying on the camera's automatic features. The histogram is a great tool to have, but it's NOT the best method to get proper exposure. At best, it's a backdoor method to get a decent picture...I just hope it wasn't a once in a lifetime shot that you missed because you didn't understand how to get the right exposure BEFORE the shutter is tripped.

47 years on film, 3 months on digital. It sure is nice to have some slight measure of the result before the film comes back. Never had that luxury on film. I use it frequently but you are right, get it right before you shoot.

[bauerman]

I can't envision myself getting to the point were I have little use of the histogram........how many different exposure situations are you going to run into in your life? - how can you know how to shoot each one of them perfectly BEFORE the shot each time? "Get it right before you shoot" for sure does sound like a film mantra and one that is not as critical in the digital age. Just my opinion.

[Todd Jacobsen]

I can't envision myself getting to the point were I have little use of the histogram........how many different exposure situations are you going to run into in your life? - how can you know how to shoot each one of them perfectly BEFORE the shot each time? "Get it right before you shoot" for sure does sound like a film mantra and one that is not as critical in the digital age. Just my opinion.

Film mantra requires a lightmeter. Old school, used reflective meter which is the DSLR. Modern school uses incident meter.

You can ALWAYS live without the histogram with an incident lightmeter...

Do you ALWAYS want to?!

[Titus213]

It is one more tool in the bag that can be used to get the good shots home.

[lancea]

You can ALWAYS live without the histogram with an incident lightmeter...
Thank goodness I hung onto my Weston Euro-Master. The real beauty of that method of exposure determination is that it means you can still set the correct exposure after the camera batteries go flat. Worked with my previous SLR anyway :mrgreen:

But I wonder if I throw away too many shots because it shows "flashing highlights". Does everyone do this, or should I be more adventurous? I'm not talking about tiny little bits of flashing highlights, but where the main subject of the picture is still a little bit dark and there is a large area flashing - usually the sky but sometimes part of the subject.

[robertwgross]

If I get even any bits of flashing highlights, then I know I'm in trouble. If it is only a little bit, then I can probably correct some of it at the RAW conversion. If it is a lot, then it is still possible to correct, but it gets unlikely to correct without a fair amount of trouble. If it was a once-in-a-lifetime shot, then careful correction might be the only possibility. Getting it correct in the camera will be a whole lot more successful than trying to correct in the computer, however.

---Bob Gross---

[PacAce]

Leo, the 10D histogram is luminance, not the green channel. I confirmed this by displaying the histogram on my 10D and by looking at the luminosity histogram on Photoshop CS. The green channel and luminance histogram often look the same, but I found one RAW file where they're very different.

I read about it on some other forum.
That's funny. Our 10Ds must behave differently because I did the same thing and my 10D was displaying histogram information that was in the green channel only. Yes, it was displaying it as a luminance info but that info came from the green channel. How do I know that? Because that image had 3 distinct and separate spikes for the red, green and blue channels in PS. The luminance historgram shows a 3-spiked histogram in PS. The 10D only shows one and that spike matches the spike from the green channel.

[PacAce]

But I wonder if I throw away too many shots because it shows "flashing highlights". Does everyone do this, or should I be more adventurous? I'm not talking about tiny little bits of flashing highlights, but where the main subject of the picture is still a little bit dark and there is a large area flashing - usually the sky but sometimes part of the subject.
If I had a camera with a very wide dynamic range, yes, I would be worried if I saw some blinking highlights on the LCD screen. However, given that the digital camera doesn't have as wide a dynamic range as one would want, I would look very critically at the blinking highlights to determine if it's an important part of the picture or not. If what's more important is in the shadows, then hang the blinking highlights and go for the shadow details. Otherwise, you'll only end up with an image that's too dark and when you try to recover the details in the shadows, you end up with noise.

On the other hand, if you're shooting a bride in full sun, then you better make darn sure you don't see any blinking highlights anywhere, except maybe for that reflection off the chrome on the car behind her. ;)

And, if both highlight and shadow details are important, then it's time to get creative with multiple exposures, graduated ND filters, etc.

The histogram and the image on the LCD are tools to help you make intelligent decisions about your exposure and that's all they are. They don't excuse you from thinking about what you are doing.

[slin100]

That's funny. Our 10Ds must behave differently because I did the same thing and my 10D was displaying histogram information that was in the green channel only. Yes, it was displaying it as a luminance info but that info came from the green channel. How do I know that? Because that image had 3 distinct and separate spikes for the red, green and blue channels in PS. The luminance historgram shows a 3-spiked histogram in PS. The 10D only shows one and that spike matches the spike from the green channel.
Are you sure you were looking at the luminance histogram in PS? The default histogram in PS is RGB which is a cumulative sum of the red, green and blue histograms. You need to change the histogram mode in PS to Luminosity to display luminance.

[slin100]

Here's an example. This is the green channel from one of my RAW files.
http://img129.echo.cx/img129/4261/green2xi.jpg

This is the luminosity.
http://img129.echo.cx/img129/8923/luminosity9cq.jpg

On my 10D, the histogram looks more like the Luminosity histogram. I really had to search my archive for an image with such disparate histograms. Most of the time the green channel histogram tracks the luminance histogram very closely.

[PacAce]

Here's an example. This is the green channel from one of my RAW files.
http://img129.echo.cx/img129/4261/green2xi.jpg

This is the luminosity.
http://img129.echo.cx/img129/8923/luminosity9cq.jpg

On my 10D, the histogram looks more like the Luminosity histogram. I really had to search my archive for an image with such disparate histograms. Most of the time the green channel histogram tracks the luminance histogram very closely.
I'm not at home so I can't show you want I have but, if I'm not mistaken, the luminosity is the combination of the 3 channels to make up the total luminosity of the image. I can't see how luminosity can be independent and different from the combination of the 3 channels. That's physically impossible but I'll double check when I get home.

[slin100]

I'm not at home so I can't show you want I have but, if I'm not mistaken, the luminosity is the combination of the 3 channels to make up the total luminosity of the image. I can't see how luminosity can be independent and different from the combination of the 3 channels. That's physically impossible but I'll double check when I get home.
Luminosity is a combination of the 3 channels but it's constructed differently from a simple cumulative RGB histogram. A luminosity histogram is calculated from the RGB triplet at each point in the image, whereas in a cumulative RGB histogram the positions of R, G and B pixels are irrelevant.

Consider these 2x2 pixel images:
(255,255,255) (000,000,000)
(000,000,000) (000,000,000)

(255,000,000) (000,255,000)
(000,000,255) (000,000,000)

These two images will have identical RGB histograms but very different luminosity histograms. The luminosity histogram of the top image will have a single spike at 255, whereas the bottom image will have three spikes. I'm not at home, either, but I believe the spikes will be around 153, 77 and 25.

[PacAce]

Luminosity is a combination of the 3 channels but it's constructed differently from a simple cumulative RGB histogram. A luminosity histogram is calculated from the RGB triplet at each point in the image, whereas in a cumulative RGB histogram the positions of R, G and B pixels are irrelevant.

Consider these 2x2 pixel images:
(255,255,255) (000,000,000)
(000,000,000) (000,000,000)

(255,000,000) (000,255,000)
(000,000,255) (000,000,000)

These two images will have identical RGB histograms but very different luminosity histograms. The luminosity histogram of the top image will have a single spike at 255, whereas the bottom image will have three spikes. I'm not at home, either, but I believe the spikes will be around 153, 77 and 25.

Yes, I agree with your there. When I said the "combination" of the 3 channels, that's exactly what I meant, the cumulative sum of each pixel point on the image. The other 3-channel display where the histograms of the three channels are displayed as histograms in their own colors are just overlays of each colored channel and not a "combination".

Now that we are both talking about the same thing, then, yes, I am very possitive that the luminosity histogram that I was looking at was different from the green channel luminosity histogram on PSCS. And the luminosity histogram on my 10D looked exactly like the luminosity histogram of just the green channel and not the histogram of the total luminosity of the image as displayed in PSCS.

[PacAce]

Consider these 2x2 pixel images:
(255,255,255) (000,000,000)
(000,000,000) (000,000,000)

(255,000,000) (000,255,000)
(000,000,255) (000,000,000)

These two images will have identical RGB histograms but very different luminosity histograms. The luminosity histogram of the top image will have a single spike at 255, whereas the bottom image will have three spikes. I'm not at home, either, but I believe the spikes will be around 153, 77 and 25.

BTW, regarding you example, I agree with you that the RGB histograms of the two will be identical and I also agree with you that the total luminosity histograms will be different. I also agree that the top will have a spike at the 255 point (3/4 of max high). But I also think that it will have a spike at the 0 point (1/4 of max high). Furthermore, I disagree that the bottom one will have three spikes. I think it will also have a spike at the 85 point (3/4 high) and another one at the 0 point (1/4 high). :)

[slin100]

BTW, regarding you example, I agree with you that the RGB histograms of the two will be identical and I also agree with you that the total luminosity histograms will be different. I also agree that the top will have a spike at the 255 point (3/4 of max high). But I also think that it will have a spike at the 0 point (1/4 of max high).

Yes, I forgot about the spike at 0.

Furthermore, I disagree that the bottom one will have three spikes. I think it will also have a spike at the 85 point (3/4 high) and another one at the 0 point (1/4 high). :)
I think you are mistaken. You are assuming that luminosity is an unweighted average of R,G, and B. Look up the conversion forumla RGB->Lab. L is the luminosity channel. L is approximately 0.30 * R + 0.59 * G + 0.11 * B.

[Edit: Oops this forumula is wrong. :o Anyway, the point is that luminosity is not a simple average of RGB]

[PacAce]

OK, I'm home and double checked the histogram I was looking at and you were correct, Steven. I was looking at the RGB histogram instead of the luminosity histogram. And the luminosity histogram does, indeed, look like the green channel, just like what I saw on my 10D. So, now I'm back to square one. I would like to see a definitve source where it says (or refutes the fact) that the 10D and similar cameras use the Green channel for the histogram displayed on the LCD.

Thanks, Steven, for clarifying this for me. :)

[slin100]

OK, I'm home and double checked the histogram I was looking at and you were correct, Steven. I was looking at the RGB histogram instead of the luminosity histogram. And the luminosity histogram does, indeed, look like the green channel, just like what I saw on my 10D. So, now I'm back to square one. I would like to see a definitve source where it says (or refutes the fact) that the 10D and similar cameras use the Green channel for the histogram displayed on the LCD.

Thanks, Steven, for clarifying this for me. :)
You are most welcome, Leo!

As to a definitive source, I haven't found one. This remark (http://www.robgalbraith.com/ubbthreads/showthreaded.php?Cat=0&Number=232421&page=0&vc=1) by Chuck Westfall confirms that the 1D2's brightness histogram is luminosity. Of course, that doesn't say anything about the 10D.

Definitive source or not, experimentation should easily confirm this. The key is to compose an image where the luminosity and green histograms are indisputably different. A contrived case where the luminosity histogram has one spike and the green histogram has two spikes should be proof enough. I'll try to capture generate a file and post it when I get home.

[PacAce]

OK, I'm 99.99% convinced that the 10D does NOT use just the green channel for the histogram display based on this test that I did.

The first image is a display of my 10D showing a test image I shot and the histogram associated with it. (Sorry, I was too lazy to set up a tripod so it's really blurry :o )

The 2nd image is the same image in PSCS2, along with the luminosity and color channel histograms. I don't think there's any doubt that the histogram displayed by the 10D very closely matches the luminosity histogram displayed in PSCS. The Green channel histogram has the same general shape but you can definitely tell that it is not what's displayed by the 10D.

[Tom W]

How about taking a picture of something solid red, then solid blue, then solid green. Compare results.

EDIT: looks like someone is already investigating this angle! :)

[PacAce]

How about taking a picture of something solid red, then solid blue, then solid green. Compare results.

EDIT: looks like someone is already investigating this angle! :)
You know what they say, Tom. Great minds think alike. ;)

[lancea]

If I get even any bits of flashing highlights, then I know I'm in trouble.
Thanks Bob and Leo. Sounds like I'm doing the right thing. I guess there must be a little bit of latitude as the highlights are presumably calculated from the same thumbnail as the histogram - also using a specific colour-channel. It was a beautiful sunny autumn day yesterday. Pretty much every shot I took had a very wide range with the histogram heavily bunched left of centre, but with significant subject-detail on the right. Quite pleased with the way they look on screen though.

Lance

[slin100]

Nice job, Leo. I guess you saved me some work! :)

This is also a great illustration of the drawback of a luminosity histogram. As one can see, the individual channels are almost blown but the luminosity histogram indicates a significant amount of headroom still available.

[Tom W]

You know what they say, Tom. Great minds think alike. ;)

Its just that some think a little sooner than others! :)

[Tom W]

Nice job, Leo. I guess you saved me some work! :)

This is also a great illustration of the drawback of a luminosity histogram. As one can see, the individual channels are almost blown but the luminosity histogram indicates a significant amount of headroom still available.

My guess FWIW is that if you increased all 3 channels by 1/3 stops, you'd get a totalized 1 stop increase on the luminosity histogram. But its only a guess.

On the other hand, if you increased the brightness of just the red channel (somehow) by 1 stop, it would be blown out while the luminosity would still be within the reasonable level. Not sure how you'd brighten up the red without the other two channels increasing as well, at least not photographically. Based on the image, any increase in exposure would affect all 3 channels simultaneously.

[rdenney]

It's really easy to see who has learned photography on a film SLR and who never learned how to expose film beyond relying on the camera's automatic features. The histogram is a great tool to have, but it's NOT the best method to get proper exposure. At best, it's a backdoor method to get a decent picture...I just hope it wasn't a once in a lifetime shot that you missed because you didn't understand how to get the right exposure BEFORE the shutter is tripped.

Though I like the sentiment of complaining about those whippersnappers with no film experience, I think I have to disagree with this statement.

In my large-format days, I was a determined user of the Zone System. I would use a 1-degree spotmeter to measure important parts of the scene, placing some on the zone scale and seeing where the others would fall. I new from having done my own calibration where those zones would end up on my film, if I followed my darkroom procedures carefully. Why did we go to this trouble? To make sure that the tones we visualized for the print would be recorded on the film. Otherwise, that visualized print would be unattainable.

We also had to worry about the pronounced toe of film's characteristic curve, where we not only had to make sure our middle values were within the range of the film, but that they were in the center of that range where the characteristic curve was steep so we could maintain tonal separation.

That was a lot of work.

Now, I can let the evaluative metering try to fit all my metering areas in the image into the dynamic range of the sensor, and then I can confirm that it did so by a quick check of the histogram. The response is quite linear, so I know if it's not clipped I can put it where I want to in Photoshop. It's much, much easier to achieve a previsualization using this technology.

In the old days, those who didn't use the Zone System were often shooting transparency film. They were interested in making a slide that would look good projected, and be an accurate reference for the printer. So, they wanted the slide dark for saturation and to protect the highlight tonal separation. But they still had to be bright enough to look good. That was a tough compromise to hit, because slide film is only good for five or six stops of scenery range. But I can get more like 10 stops of scenery range on my 10D, and much of the time that accommodates the scene completely. Thus, I don't have to make that compromise at exposure time.

Film was so limiting, and if we feel compelled to expose as carefully as we used to, we are constraining ourselves with the narrow capabilities of film compared to the digital sensors. Adams would never have come up with the Zone System if he'd had histograms available to him, but he still would have taught the importance of previsualization.

Rick "happy to be spending that time understanding the scene rather than fidding with that Pentax Spotmeter" Denney

[Tom W]

I confess to using the little chart that came with every roll of Kodak film to help determine exposure on occasion. That and the center-weighted average light meter on the FT. Never had a separate meter, but the camera's meter did a very good job. Sometimes, a little intuitive over- or under- exposure was desirable, but one kind-of got a feel for that sort of thing (or at least one that didn't have a separate light meter).

[PacAce]

Nice job, Leo. I guess you saved me some work! :)

This is also a great illustration of the drawback of a luminosity histogram. As one can see, the individual channels are almost blown but the luminosity histogram indicates a significant amount of headroom still available.
And that's why I'm so glad that I have the ability to set my 1DmkII to display the histogram as 3 separate RGB histograms instead of a single luminosity histogram. :)