PDA

View Full Version : Don't rely on histograms when shooting RAW?


memmons
23rd of October 2009 (Fri), 18:34
So, I'm reading Martin Evening's "The Adobe Photoshop Lightroom 2" book and came across this passage, which surprised me:

If you are shooting raw, it is unwise to place too much emphasis on the camera histogram. It is best to either trust the exposure system in the camera to get it right or rely on the histogram in Lightroom.

For those still learning, such as myself, not being able to use the on-camera histogram to get a better idea what may or may not be a well exposed shot would be a huge drawback of shooting RAW. Additionally, I would prefer not to always shoot in auto mode. This bothered me enough that I started doing some research and learned something quite helpful--which I'm sure is obvious to old pros, but it wasn't pointed out in the several books I've read:

The on-camera histogram is not based on the RAW data, but rather the converted jpeg (or other format) which is shown on the camera display. This leads to the discrepancy which Martin pointed out. However, you can adjust your camera so the histogram does closely reflect the RAW data and what will be seen in PP software such as Lightroom. To do this, make sure your color mode is set to Adobe RGB and adjust the camera's contrast, tint and saturation setting to minimum when shooting RAW.

I'd love to get some feedback from more experience POTN members that this is indeed correct and if I'm missing anything in regards to getting an accurate histogram representation when shooting RAW.

Thanks!

RDKirk
23rd of October 2009 (Fri), 18:51
The camera meter is an abstract interpretation of the conditions that may exist when you take the exposure at some point in the future.

The histogram is a report of what actually happened in the sensor when you made the exposure.

Both require learning and experience to interpret correctly, but a correct interpretation of the report of actually will be more accurate than a correct interpretation of a prediction of conditions. It's like the difference between last week's prediction of this morning's rain and this afternoon's report of this morning's rain.

focus.pocus
23rd of October 2009 (Fri), 18:55
wait a sec. You shoot RAW but shoot in auto mode??? you're kidding right???

number six
23rd of October 2009 (Fri), 18:57
Check out this thread: http://photography-on-the.net/forum/showthread.php?t=485349

JeffreyG
23rd of October 2009 (Fri), 19:05
I'd love to get some feedback from more experience POTN members that this is indeed correct and if I'm missing anything in regards to getting an accurate histogram representation when shooting RAW.

Thanks!

I don't agree with a lot of what was stated in the quote.

For starters, it is worthwhile to understand that the camera histogram is based on a thumbnail jpeg, and if you have any serious modifications set up (picture styles etc) then it can be distorted. But if you shoot raw and leave the picture style set to 'neutral' the histogram will be close. It won't show you just how much headroom can be regained in post etc, but it's going to give you a very good idea.

The author looses all credibility in my opinion when he suggests it's better to trust the camera light meter than the histogram. The light meter will generally be all over the plavce depending on the scene tonality. I almost never use the meter in my cameras, I use incident light meters and the histogram almost exclusively.

tzalman
24th of October 2009 (Sat), 09:28
http://photography-on-the.net/forum/showthread.php?p=8846648#post8846648

memmons
24th of October 2009 (Sat), 11:33
wait a sec. You shoot RAW but shoot in auto mode??? you're kidding right???

I did not say full-auto mode...

memmons
24th of October 2009 (Sat), 11:33
Check out this thread: http://photography-on-the.net/forum/showthread.php?t=485349

That thread is excellent. Thank you!

memmons
24th of October 2009 (Sat), 11:37
The author looses all credibility in my opinion when he suggests it's better to trust the camera light meter than the histogram. The light meter will generally be all over the plavce depending on the scene tonality. I almost never use the meter in my cameras, I use incident light meters and the histogram almost exclusively.

Thanks for the good info. While I do think Martin has some good information to impart, I am not a fan of his writing style.

JeffreyG
24th of October 2009 (Sat), 12:34
Thanks for the good info. While I do think Martin has some good information to impart, I am not a fan of his writing style.

He would have been safe if he was simply pointing out that care needs to be taken when consulting the histogram considering jpeg conversion, white balance and RAW headroom.

But to imply that one is better off bailing on the histogram to consult a reflective meter is nutty.

airfrogusmc
24th of October 2009 (Sat), 12:45
If your shooting a polar bear in snow the histogram is going to look much different and should if you EXPOSED PROPERLY than if you were shooting a subject in all dark clothing on a dark background. I would suggest researching why.

RDKirk
24th of October 2009 (Sat), 13:27
If your shooting a polar bear in snow the histogram is going to look much different and should if you EXPOSED PROPERLY than if you were shooting a subject in all dark clothing on a dark background. I would suggest researching why.

The histogram is a report of what the sensor is actually collecting, while the meter reports a quantity of reflected light or incident light with a simple sliderule function that changes the reading into camera settings. The histogram--much closer to the actuality of the collected image--is potentially the very most accurate source of exposure information available.

But it definitely does take knowledge and experience to exploit it effectively.

JWright
24th of October 2009 (Sat), 14:16
The camera has a histogram? What will they think of next?

Seriously, I never rely on my camera's histogram. I do use the histogram in LR a bit, but just to get a handle on blown highlights or blocked shadows. Most of the time I rely on the camera meter to get the exposure reasonably correct. Since I shoot exclusively in RAW, I have the ability to make corrections in PP.

CyberDyneSystems
24th of October 2009 (Sat), 14:21
I don't agree with a lot of what was stated in the quote.
...

Ditto.

The camera's Histogram is very important, and the best way for us to be aware of the exposure we are making.
....and how on earth is the Histogram in lightroom to be relied on when out shooting in the field? I believe either this was a silly statemenet which we should discount all together, or more likely it was a fragment that does not tell the authors full meaning.

kingdaddy
24th of October 2009 (Sat), 14:57
How can you see the histogram without going into "Live Mode"?

airfrogusmc
24th of October 2009 (Sat), 15:03
Ditto.

The camera's Histogram is very important, and the best way for us to be aware of the exposure we are making.
....and how on earth is the Histogram in lightroom to be relied on when out shooting in the field? I believe either this was a silly statemenet which we should discount all together, or more likely it was a fragment that does not tell the authors full meaning.

AGREE. I feel its a very important tool and any photographer that wants to get really consistent images should take some time to learn how to read it. Like in my example its usually much more than having all the tones fall evenly its sometime about where you want the tones to fall.

joedlh
24th of October 2009 (Sat), 15:05
wait a sec. You shoot RAW but shoot in auto mode???

I don't see that one is related to the other. There are some settings where some variation of auto mode is entirely appropriate. Whether one shoots jpegs or raw images in that setting is irrelevant and the creative decision of the photographer.

joedlh
24th of October 2009 (Sat), 15:20
How can you see the histogram without going into "Live Mode"?

In the xxD cameras, while an image is displayed, press the Info button on the back of the camera until it comes up.

It has been noted that one must understand how to use the histogram. But there were no examples provided. Here are a couple: We alwasys try to reproduce the tonal "character" of the setting. For example, somebody mentioned the polar bear on white snow. Your camera's meter will try to make all that white 18% gray. The histogram might show a peak in the middle, which is not appropriate. Rather, you would want to see a peak on the right side to indicate that the whites will be presented as white. Similarly, at a concert, there's lots of unlit portions of the stage. To represent the perceived tonal range of a dark stage with lit performers, you would want to see in the histogram a lump on the left side (black) with a peak for the lit performers somewhere to the right of middle, but not all the way to the right unless they are wearing white face paint, dressed in white clothes, playing white instruments, and the light shining on them is white.

joedlh
24th of October 2009 (Sat), 15:29
The on-camera histogram is not based on the RAW data, but rather the converted jpeg (or other format) which is shown on the camera display.

Can somebody provide a Canon-issued source for this information? I would be quite disappointed if when shooting in raw mode my camera was showing me what the light distribution would look like if I was in fact shooting jpegs. As a matter of fact, my experience with using the histogram in raw mode is quite close to what the actual raw image looks like. For example, it's showing what I would expect from a 14-bit per color channel tonal range rather than the 8-bit per channel jpeg.

tzalman
24th of October 2009 (Sat), 18:03
How can you see the histogram without going into "Live Mode"?
By seeing it in review and reshooting if neccessary.

Wilt
24th of October 2009 (Sat), 19:01
My own interpretation of Evening's statement, "If you are shooting raw, it is unwise to place too much emphasis on the camera histogram. It is best to either trust the exposure system in the camera to get it right or rely on the histogram in Lightroom", (even if it is an extracted fragment of a larger explaination) would be that the camera histogram can be misleading compared to the histogram in RAW conversion programs, expecially when one understands that RAW convertors can save information that would be lost within JPG file alone. So one should use the camera histogram as an initial guide, but know that more data can be recovered via RAW conversion. Of course, if you shoot only JPG, my own interpretation would not apply. But the Evening book is about Lightoom 2, a RAW convertor!

RDKirk
24th of October 2009 (Sat), 19:27
Can somebody provide a Canon-issued source for this information? I would be quite disappointed if when shooting in raw mode my camera was showing me what the light distribution would look like if I was in fact shooting jpegs. As a matter of fact, my experience with using the histogram in raw mode is quite close to what the actual raw image looks like. For example, it's showing what I would expect from a 14-bit per color channel tonal range rather than the 8-bit per channel jpeg.

The statement is true enough, but the light distribution is the same regardless where you're processing the raw (in camera or in computer). Processing in the computer gives you access to a bit greater dynamic range--you won't be clipping either end as much. But the distribution is the same.

RDKirk
24th of October 2009 (Sat), 19:33
In the xxD cameras, while an image is displayed, press the Info button on the back of the camera until it comes up.

It has been noted that one must understand how to use the histogram. But there were no examples provided.

Another use of the histogram is to make sure a specific important tone is being placed on the sensor's range where we want it. That's what we did as part of the Zone System, and we can do that with the histogram.

With film, the tone considered most important to place properly was middle gray (Zone V), which must be placed at the center of the film's dynamic range. Film's dynamic range is variable by development--the middle is the anchor point.

On the digital sensor, the anchor point is at the top--the point highlight tones hit the wall. Therefore, the most important tone to place properly is the brightest tone that must retain detail. If you place that tone just under the top of the sensor's range, you know that "brightest tone that must retain detail" will retain its detail and you've given the scene the fullest exposure to pull as much shadow detail above base noise as possible. You can do that precisely with the histogram.

kingdaddy
24th of October 2009 (Sat), 20:32
In the xxD cameras, while an image is displayed, press the Info button on the back of the camera until it comes up.
.

I dont have any "INFO" button on any SLR canon camera I own.

DStanic
24th of October 2009 (Sat), 22:04
I dont have any "INFO" button on any SLR canon camera I own.

INFO on my 30D
DISP on my XTi

same thing.

argyle
25th of October 2009 (Sun), 07:37
Expose to the right of your histogram, making sure not to slam up against the right side. This will retain detail in the highlights and keep you from blowing them out. Also, if your camera has this feature, set your histogram to RGB mode...this will show you the range of each channel. If you are on the verge of blowing any highlights in either channel, (histogram slammed up against the far right side), simply adjust your exposure compensation dial to bring it down a bit. Or, if any of the channels is well below the right side, adjust your Ec in the opposite direction to bring the exposure more to the right (again, without slamming up against the right side of the histogram).

rral22
25th of October 2009 (Sun), 09:26
A long time ago when the earth was green, I was shooting slides for stock. The mantra was, "Expose for the highlights, let the shadows fall where they may."

Digital seems the same to me, and I use the histogram very often as a tool to be sure I have not blown anything beyond rescue. I find the histogram, both in camera and in post, to be one of the most useful tools I have for getting exposures the way I want them.

kingdaddy
25th of October 2009 (Sun), 09:33
What good does it do to look at the histogram after you take the shot, by then you already know if it is exposed properly by looking at the picture, plus many shots are a one time thing.

Sorry but I just don’t get the histogram thing unless you are using live view which is cumbersome. If you can’t see it before you take the shot I personally have no use for it.

JeffreyG
25th of October 2009 (Sun), 09:35
What good does it do to look at the histogram after you take the shot, by then you already know if it is exposed properly by looking at the picture, plus many shots are a one time thing.

Sorry but I just don’t get the histogram thing unless you are using live view which is cumbersome. If you can’t see it before you take the shot I personally have no use for it.

Just one example? I recently shot a volleyball tournament. I took one shot based on my Sekonic and then I checked the histogram. I adjust 1/3 stop based on the histogram to ETTR.

And then I took the next 680 shots through the entire day without reviewing any image, histogram or light meter.

That's how I use the histogram.

DStanic
25th of October 2009 (Sun), 09:55
What good does it do to look at the histogram after you take the shot, by then you already know if it is exposed properly by looking at the picture, plus many shots are a one time thing.



There has been lots of times I thought I got a "perfect" exposure when reviewing on the LCD, but after when I look at them on the computer they are too dark. LCDs are not accurate for judging exposure.

That's why you take a test shot or 2, use the histogram and adjust for how the real shots will turn out. It's too bad there is not a "live" histogram (without using live view of course).

CyberDyneSystems
25th of October 2009 (Sun), 10:00
... Also, if your camera has this feature, set your histogram to RGB mode...this will show you the range of each channel. If you are on the verge of blowing any highlights in either channel, (histogram slammed up against the far right side), simply adjust your exposure compensation dial to bring it down a bit. Or, if any of the channels is well below the right side, adjust your Ec in the opposite direction to bring the exposure more to the right (again, without slamming up against the right side of the histogram).

I wonder of the author would have said the same thing if they were discussing RGB histograms.. ?

All my cameras for some time have had this feature, and have been set to RGB histogram,. so I forgot it's "not the norm"

It doesn't alter my stance re ANY in camera histogram being a valuable tool, but I will agree that the RGB Histogram is the better of the two.

CyberDyneSystems
25th of October 2009 (Sun), 10:03
What good does it do to look at the histogram after you take the shot, by then you already know if it is exposed properly by looking at the picture, ...

I don't think looking at a tiny pic on my camera's LCD gives me 1/10th the info I get from looking at a Histogram.

Also many of us "got to know our histograms" from back in the day when the on camera LCD was virtually useless for seeing things like exposure and sharpness.. now things have improved dramatically, but still the Histo info offers vastly more information to the shooter.

Your statement I think reflects a lack of understanding of the histogram,. which is fair, but hardly the sort of wisdom that should be reinforced.

I am leary of any advice that tells us to ignore a significant and important tool.
We see a lot of this still, but saw more of it as the progression to digital was underway from Film.

Use the tools at your disposal. If you are ignoring some of them, you simply have more room to grow.

tdodd
25th of October 2009 (Sun), 10:07
Can somebody provide a Canon-issued source for this information? I would be quite disappointed if when shooting in raw mode my camera was showing me what the light distribution would look like if I was in fact shooting jpegs. As a matter of fact, my experience with using the histogram in raw mode is quite close to what the actual raw image looks like. For example, it's showing what I would expect from a 14-bit per color channel tonal range rather than the 8-bit per channel jpeg.
Take a look at this video, which shows how the histogram changes when the white balance is adjusted....

http://www.youtube.com/user/eezytiger#p/u/5/FLXoznsvEvI

Why would the histogram change if it was reflecting the raw captured data? It shouldn't. The problem is that in camera histograms are generated once the camera has effed about with the data to produce a JPEG with all the attendant manipulation to convert to a colour space, colour balance, sharpen, add a tone curve etc. etc..

RDKirk
25th of October 2009 (Sun), 10:37
What good does it do to look at the histogram after you take the shot, by then you already know if it is exposed properly by looking at the picture, plus many shots are a one time thing.

Exposure rarely changes rapidly, unpredictably, and unrepeatedly at the same time. Exposures are almost always either constant, predictable, or repeatable.

RDKirk
25th of October 2009 (Sun), 10:39
I am leary of any advice that tells us to ignore a significant and important tool.
We see a lot of this still, but saw more of it as the progression to digital was underway from Film.

Especially since the histogram is giving us what the military calls "ground truth" of what the sensor actually did rather than a distant representation of what may happen.

rral22
25th of October 2009 (Sun), 11:48
What good does it do to look at the histogram after you take the shot, by then you already know if it is exposed properly by looking at the picture, plus many shots are a one time thing.


The LCD is, in my opinion, the LEAST reliable measure of exposure on the whole camera.

While many shots are a ""one time thing, using the histogram for those that (as some have pointed out) it is a very easy to use and at least reasonably scientific measure makes a lot of sense.

It is a very useful tool. To refuse to use it when it is the proper tool is just silly.

DStanic
25th of October 2009 (Sun), 12:39
Check out this link on UniWB
http://www.guillermoluijk.com/tutorial/uniwb/index_en.htm

number six
25th of October 2009 (Sun), 13:02
I dont have any "INFO" button on any SLR canon camera I own.

Dunno about your XSi, but on your 300D it's just below the "menu" button.

-js

Wilt
25th of October 2009 (Sun), 19:33
Take a look at this video, which shows how the histogram changes when the white balance is adjusted....

http://www.youtube.com/user/eezytiger#p/u/5/FLXoznsvEvI

Why would the histogram change if it was reflecting the raw captured data? It shouldn't. The problem is that in camera histograms are generated once the camera has effed about with the data to produce a JPEG with all the attendant manipulation to convert to a colour space, colour balance, sharpen, add a tone curve etc. etc..

Excellent demonstration that the histogram is affected by the WB setting, even when the camera is set for RAW and the light is not at all altered in color balance -- which should result in a uniform histogram regardless of WB setting if it indeed reflected the light striking the sensor, but obviously it is being affected by in-camera processing!

And given the fact that the JPG data used for histogram display in camera is a decimated representation of the fuller pixel count of the sensor, you might well have many pixels in the actual full resolution image which are clipped in the highlights, yet never see them because those same pixels could be missing from the fewer pixels analyzed in the histogram. So reliance upon in-camera histogram is not assurance that data is not being clipped, after all! It is a close approximation, at best.

rral22
25th of October 2009 (Sun), 20:56
So reliance upon in-camera histogram is not assurance that data is not being clipped, after all! It is a close approximation, at best.

No argument. But, since all exposure calculations from in camera meters have ALWAYS been a close approximation at best, the histogram is a very useful approximation, especially if the photographer understands what it is saying.

All camera meters have always required the photographer to understand just what any particular reading meant and how to use that measure to get what was really wanted. Histograms are, with all their foibles, very nice tools to help get what is needed.

And "what is needed" is the key. Negatives or Raw files have always been shot with the plan to post process for what the photographer intended, not always as a "final product". Push processing of film, dodging and burning in the dark room, applying curves in photoshop are all the same processes, and photographers always used them, and shot the film or digital file with the intention of giving it the qualities necessary for those manipulations (the main purpose of ETTR for example).

To be critical of a histogram because it is not the perfect representation of what you think it should be for any given shot is just not reasonable, and does not make the tool useless by any means.

memmons
25th of October 2009 (Sun), 21:17
I believe either this was a silly statemenet which we should discount all together, or more likely it was a fragment that does not tell the authors full meaning.

It wasn't taken out of context. That was the complete statement with no other qualifiers after it. Page 257 of the book..image attached. But in any case, it's clear that the statement was misleading. This thread cleared it up for me and I learn quite a bit in the process.

402697

memmons
25th of October 2009 (Sun), 21:40
Check out this link on UniWB
http://www.guillermoluijk.com/tutorial/uniwb/index_en.htm

Thanks for that! Good stuff. Did anyone else find it ironic the photographer in the demo, who obviously knows a lot about photography, didn't turn the video cam's auto-focus off so it could focus properly? :lol:

Wilt
25th of October 2009 (Sun), 21:52
No argument. But, since all exposure calculations from in camera meters have ALWAYS been a close approximation at best, the histogram is a very useful approximation, especially if the photographer understands what it is saying.

All camera meters have always required the photographer to understand just what any particular reading meant and how to use that measure to get what was really wanted. Histograms are, with all their foibles, very nice tools to help get what is needed.

And "what is needed" is the key. Negatives or Raw files have always been shot with the plan to post process for what the photographer intended, not always as a "final product". Push processing of film, dodging and burning in the dark room, applying curves in photoshop are all the same processes, and photographers always used them, and shot the film or digital file with the intention of giving it the qualities necessary for those manipulations (the main purpose of ETTR for example).

To be critical of a histogram because it is not the perfect representation of what you think it should be for any given shot is just not reasonable, and does not make the tool useless by any means.

Wait, I am NOT critical of histograms, I merely state what the shortcomings (and strengths) of the tools we use, so we the users are not fooled into false security!!! And give the OP and the quotation that he asked for interpretation, my comment was in agreement with the quoted statement (and I have no argument against the quoted statement)

RDKirk
25th of October 2009 (Sun), 21:56
Take a look at this video, which shows how the histogram changes when the white balance is adjusted....

http://www.youtube.com/user/eezytiger#p/u/5/FLXoznsvEvI

Why would the histogram change if it was reflecting the raw captured data? It shouldn't. The problem is that in camera histograms are generated once the camera has effed about with the data to produce a JPEG with all the attendant manipulation to convert to a colour space, colour balance, sharpen, add a tone curve etc. etc..

This doesn't show anything we didn't already know. We already know the histogram has been developed according to the JPEG settings. We already know how channels are boosted to compensate for the lack of parts of the spectrum that vary from what we see as white light. To get that back into balance on the histogram--just set the white balance on the camera properly (close is good enough)...as the video clearly showed. We'd also want to "neutralize" the other JPEG settings as much as possible. We already know this stuff.

Failing to understand these simple things is like not knowing a reflected light meter is always calculating an exposure for middle gray. This goes into what I've said before about knowing how to properly interpret the histogram.

Curtis N
25th of October 2009 (Sun), 21:58
But to imply that one is better off bailing on the histogram to consult a reflective meter is nutty.Agreed.
Exposure rarely changes rapidly, unpredictably, and unrepeatedly at the same time. Exposures are almost always either constant, predictable, or repeatable.Unless you're shooting theatre. ;)

The camera histogram is a tool. It's not meant to replace other tools, it's meant to add to them. Like any tool, you need to learn how to use it. And understand that:
1) It's based on the imbedded JPEG, which is based on your camera settings for white balance, contrast and saturation.
2) It's a small, graphical representation of millions of pixels. It won't illustrate small areas of blown highlights.

RDKirk
25th of October 2009 (Sun), 22:04
Unless you're shooting theatre.

Theatrical light schemes are planned in advance. The gaffer isn't back there playing eeny-meeny-miny-moe with the lights. It doesn't take long to figure out the settings--I did that even using hand-held meters with Mamiya press cameras back in the day shooting theater.

People try to make digital so hard, but it's so bloody easy, it's criminal. Shooting night football in snow with a press camera and a 12-pound Graflex flash is hard.

Wilt
25th of October 2009 (Sun), 22:06
Theatrical light schemes are planned in advance. The gaffer isn't back there playing eeny-meeny-miny-moe with the lights.


But as the photographer, we are not privy to the scripting, to to us it can be ever-changing! :)

And Curtis, you forgot about shooting in broken clouds when the wind is moving them quickly through the sky so that the lighting is perpetually fluctuating on the scene! :cool: