Exposure compensation

[vorlich]

Hi All,

I'm confused about exposure compensation. I understand that the metering in most cameras will try to expose bright scenes as middle grey, snowy scenes for example. However, what I don't understand is the benefit of using exposure compensation over say, taking a shot, checking the histogram and adjusting the exposure manually by changing the shutter speed to taste. This seems much less fiddly than going into a menu and making the adjustment there.

Am I missing something?

[francois]

… However, what I don't understand is the benefit of using exposure compensation over say, taking a shot, checking the histogram and adjusting the exposure manually by changing the shutter speed to taste. This seems much less fiddly than going into a menu and making the adjustment there.
…

On most DSLRs, changing exposure compensation is just a matter of turning a wheel or pressing a button. It's not as flexible as controlling exposure via the manual mode (M) but it's quick and you can leave your camera in Av or Tv mode.

[vorlich]

On most DSLRs, changing exposure compensation is just a matter of turning a wheel or pressing a button. It's not as flexible as controlling exposure via the manual mode (M) but it's quick and you can leave your camera in Av or Tv mode.

Ah-ha. Thanks for the speedy response. It makes sense now, I always shoot manual, so I hadn't considered it's use for Av or Tv modes.

[francois]

Ah-ha. Thanks for the speedy response. It makes sense now, I always shoot manual, so I hadn't considered it's use for Av or Tv modes.

Exposure compensation means compensate for the automatic meter. If you use manual mode, then you don't need to compensate - yourself -…  

[digitaldog]

However, what I don't understand is the benefit of using exposure compensation over say, taking a shot, checking the histogram and adjusting the exposure manually by changing the shutter speed to taste. T

Am I missing something?

Well for one, if you're shooting Raw, that histogram is a big fat lie. Its a histogram representing a JPEG if you had the camera set to provide one.

[vorlich]

Well for one, if you're shooting Raw, that histogram is a big fat lie. Its a histogram representing a JPEG if you had the camera set to provide one.

Yes, I'm aware of that. But it's a pretty good guide IME. If in doubt I usually bracket...

[digitaldog]

Yes, I'm aware of that. But it's a pretty good guide IME.

I totally disagree, its not at all even a close guide to the Raw data. You're looking at a baked, gamma corrected, white balanced histogram. The Raw histogram is very, very different. Be useful if we could see it.

[vorlich]

I totally disagree, its not at all even a close guide to the Raw data. You're looking at a baked, gamma corrected, white balanced histogram. The Raw histogram is very, very different. Be useful if we could see it.

Of course the RAW histogram would be preferable, but in it's absence, what do you suggest?

[digitaldog]

Of course the RAW histogram would be preferable, but in it's absence, what do you suggest?

http://www.digitalphotopro.com/technique/c...ng-for-raw.html

My suggestion is to totally disregard the LCD histogram for anything other than focus, composition, etc. IOW, the histogram is just a totally unreliable indicator for exposure unless you're exposing for the JPEG.

[vorlich]

http://www.digitalphotopro.com/technique/c...ng-for-raw.html

My suggestion is to totally disregard the LCD histogram for anything other than focus, composition, etc. IOW, the histogram is just a totally unreliable indicator for exposure unless you're exposing for the JPEG.

For those of us who meter in-camera, I'd argue the histogram is still a useful guide, provided the shooter is aware of it's limitations.

I make an exposure, ETTR, and see the histogram shows I'm just clipping the highlights, surely this is good enough, as I have left myself some headroom in the RAW capture. Bear in mind, I'm using an entry level dSLR, which I assume will be closer to the JPG histogram than your 5D. Isn't knowing the characteristics of your camera and how much headroom exists over what the histogram shows, so you can predict the correct exposure, a viable solution in lieu of a RAW histogram?

[digitaldog]

Bear in mind, I'm using an entry level dSLR, which I assume will be closer to the JPG histogram than your 5D.

I can't imagine why that would be so.

Isn't knowing the characteristics of your camera and how much headroom exists over what the histogram shows, so you can predict the correct exposure, a viable solution in lieu of a RAW histogram?

Again, the JPEG is a baked, gamma corrected, rendered image from the Raw. The histogram is based on that rendering. The Raw histogram if we could see one, would look vastly different. And if we took that Raw into any Raw converter, ever differing setting applied to produce a rendering would produce a differing histogram. Yet the original source histogram in all those examples is the same.

If I "over expose" 1.5 stops past what my meter tells me is "correct", at least on my 5D, I'll see a blown out histogram on the LCD, I'll also see a blow out histogram using a default rendering in say Lightroom. But I can move the Exposure slider such that I "normalize" (using Michael's term) the rendering such that no highlight data is blown out and the resulting histogram looks fine. The histogram on the camera represents just one fixed rendering. Within reason, I can alter that by a huge degree in a Raw converter yet the original source data is what it is.

Histograms are not future predictors. They tell us the data distribution based on some current settings. So I don't see how the one baked rendering of a JPEG I'm not producing is useful in gauging the actual data of the Raw.

[marcmccalmont]

I can't imagine why that would be so.



Again, the JPEG is a baked, gamma corrected, rendered image from the Raw. The histogram is based on that rendering. The Raw histogram if we could see one, would look vastly different. And if we took that Raw into any Raw converter, ever differing setting applied to produce a rendering would produce a differing histogram. Yet the original source histogram in all those examples is the same.

If I "over expose" 1.5 stops past what my meter tells me is "correct", at least on my 5D, I'll see a blown out histogram on the LCD, I'll also see a blow out histogram using a default rendering in say Lightroom. But I can move the Exposure slider such that I "normalize" (using Michael's term) the rendering such that no highlight data is blown out and the resulting histogram looks fine. The histogram on the camera represents just one fixed rendering. Within reason, I can alter that by a huge degree in a Raw converter yet the original source data is what it is.

Histograms are not future predictors. They tell us the data distribution based on some current settings. So I don't see how the one baked rendering of a JPEG I'm not producing is useful in gauging the actual data of the Raw.

The $64,000 question then is how do you consistently get proper ettr exposures without reference to the histogram? My experiance on several cameras is to set the histogram to RGB and the display to blink when clipped. I consistently get good ettr's with mild blinking and a small amount of clipping on the histogram. Just my opinion, the histogram is an invaluable tool for exposure but would be even better if you could select a RAW histogram
Marc

[Guillermo Luijk]

Isn't knowing the characteristics of your camera and how much headroom exists over what the histogram shows, so you can predict the correct exposure, a viable solution in lieu of a RAW histogram?

If you are really interested in the RAW histogram, why should you put any effort to find out how to _predict_ the correct exposure from a processed RAW, i.e. the JPEG? why not complain at the camera manufacturers that (in a wrong way) think you should not be interested in the RAW histogram?

It's amazing to see how many RAW shooters are not only not unhappy with the JPEG histograms their cameras provide, but glad to see thay can 'recover' information when putting the RAW file into a RAW developer. Come on, complain at those camera vendors that live on the money you put in their products, and don't be so happy trying to predict what the right exposure achieved was, just because some brilliant engineer (or more likely some marketer) that will only take weekend family pictures thought you are not intelligent enough to need or be interested in RAW histograms.

BR

[digitaldog]

The $64,000 question then is how do you consistently get proper ettr exposures without reference to the histogram?

If I could copy and paste from another forum, the reply to that question (it boils down to, it depends):

I never shoot JPEG. I never pay any attention to the Histogram or clipping.

If I have control of the situation I'm shooting in, certainly a setup where I'm lighting the scene and have time, I'll implement ETTR based on what I know about the sensor.

I also shoot a lot "from the hip" (just friends at an event, having dinner or something). Usually low light. ETTR just doesn't work as I'm always wanting either higher shutter speeds or depending on the lens, its not fast enough to lose that extra 1.5 stops. The preferred lens I use in these situations is a 35mm 1.4 but more often I'm just taking along one lens, a 24-104mm F4. The new 5DMII certainly helps being at ISO 3200 in some cases. A sharp capture is always preferable to one that isn't even if there's less noise (duh).

If the light is changing a lot and again, I'm shooting quickly and have little control, ETTR isn't going to fly. I don't want to clip highlights! I'd rather have a "normal" meter exposure with more noise in the shadows as opposed to really clipping highlights. I can't get that data back, I can deal with noise.

Ultimately, whenever the situation allows, I bracket! There's never any downside there. That would mean something that allows the time to do so without losing a shot (so no, I would never bracket shooting people or something moving that might result in the better shot being poorly exposed).

Sometimes I'm using an external meter (a Sekonic that has both spot and incident capabilities) but quite often, I am traveling light and use the camera meter.

Again, it would be really useful if the feedback on the LCD were based on what I'm actually shooting for, the Raw data. I'm not sure why the camera manufactures don't provide this option, seems easy to do.

[bjanes]

I totally disagree, its not at all even a close guide to the Raw data. You're looking at a baked, gamma corrected, white balanced histogram. The Raw histogram is very, very different. Be useful if we could see it.

A totally raw histogram without gamma encoding or logarithmic encoding would not be that useful, since (as Bruce Fraser and others have noted) the data would be scrunched up on the left side of the histogram. It is true that the camera histogram represents data from the JPEG preview and is affected by the color space and tone curve of the camera settings. However, if one uses a normal or low contrast tone curve, the camera histogram often gives a reasonably accurate indication of exposure to the right (ETTR). The tone curve affects chiefly the mid-tones and has minimal effect on the highlights. I have done experiments with my Nikon D3 and have found that the camera histogram is slightly conservative, but gives a good approximation of ETTR. In his seminal article on ETTR, Michael Reichman recommended using the camera histogram and that is what most photographers do.

I posted my results in the Adobe ACR Forum. Unfortunately, the illustration links are no longer valid, but the text is still there.

IHMO, an ideal camera histogram would reflect raw data with log base 2 encoding so that each division on the histogram would represent 1 f/stop. Lacking such an ideal, the current histograms give valuable data if one has done tests and knows how to interpret them.

[digitaldog]

A totally raw histogram without gamma encoding or logarithmic encoding would not be that useful, since (as Bruce Fraser and others have noted) the data would be scrunched up on the left side of the histogram.

I don't recall Bruce saying it would not be useful, can you find a reference? Yes, it would be all scrunched up to the left, especially if plotted like a gamma corrected Histogram. And yes, users would need to learn to read a histogram that way. And yes, the manufacturers could do something useful like put some line or reference to the right of this scrunched up histogram that indicated based on some testing, where full sensor exposure overload happened so we could adjust to just shy of that point. And yes, that would be real useful IMHO!

It is true that the camera histogram represents data from the JPEG preview and is affected by the color space and tone curve of the camera settings. However, if one uses a normal or low contrast tone curve, the camera histogram often gives a reasonably accurate indication of exposure to the right (ETTR).

Accurate by how much? You have some data to share? 1/10 a stop? 1/2? I've fooled around with my Canon's and 1.5 stops is more like it. Now maybe you consider that an accurate amount to extrapolate but I'd like a bit more precision.

I posted my results in the Adobe ACR Forum. Unfortunately, the illustration links are no longer valid, but the text is still there.

I posted mine here:

http://www.digitalphotopro.com/technique/c...ng-for-raw.html

Maybe Canon's are way, way different from Nikon.

[Panopeeper]

A totally raw histogram without gamma encoding or logarithmic encoding would not be that useful, since (as Bruce Fraser and others have noted) the data would be scrunched up on the left side of the histogram

This is nonsense, no matter who noted it. The histogram is very lop-sided if the shot was underexposed by several stops. If the shot is underexposed by one stop, then already half of the histogram is occupied.

A log-based histogram would be better, but we don't have it.

The gamma encoding is very disturbing in ETTR, when judging how much exposure can be added without clipping: the histogram appears close to the right edge - but how much is the distance in terms of EV?

Accurate by how much? You have some data to share? 1/10 a stop? 1/2? I've fooled around with my Canon's and 1.5 stops is more like it

The flashing clipping indication with my 40D is accurate within the adjustibility, i.e. 1/3 EV (the histogram is accurate as well, but not as useful). However, the price is, that all my shots are greenish, i.e. the preview JPEG and the thumbnails are good for nothing but identifying the shot, and I always have to (would have to) think of an extra shot for WB if the scenery itself does not offer anything for picking WB.

[digitaldog]

The flashing clipping indication with my 40D is accurate within the adjustibility, i.e. 1/3 EV (the histogram is accurate as well, but not as useful). However, the price is, that all my shots are greenish, i.e. the preview JPEG and the thumbnails are good for nothing but identifying the shot, and I always have to (would have to) think of an extra shot for WB if the scenery itself does not offer anything for picking WB.

Yup, that's also a severe issue that needs fixing. The LCD previews are so darn useful (for a guy from the old Polaroid days), its a shame those shooting something that isn't a JPEG get penalized. I keep thinking the camera manufacturers will get it some day. But until they figure we're all tired of larger and larger chips to get us to upgrade, I don't hold out much hope.

[vorlich]

Histograms are not future predictors. They tell us the data distribution based on some current settings. So I don't see how the one baked rendering of a JPEG I'm not producing is useful in gauging the actual data of the Raw.

Yes, but if your in-camera histogram is bunched up at the extreme right, your default histogram in the RAW converter will be similar. Of course this can be adjusted or normalised as required (within limits), but the histogram still serves as an indicator (maybe not a wholly accurate one) of where the values have been recorded in the RAW file. I'm surprised you're unable to make this connection.

I won't be abandoning the in-camera histogram any time soon...

[Panopeeper]

Yes, but if your in-camera histogram is bunched up at the extreme right, your default histogram in the RAW converter will be similar. Of course this can be adjusted or normalised as required (within limits), but the histogram still serves as an indicator (maybe not a wholly accurate one) of where the values have been recorded in the RAW file

1. If your in-camera histogram is bunched up at the extreme right, then your shot is probably worthless anyway.

2. The default histogram in the raw processor will be similar to that displayed in-camera if you are using a raw processor, which applies the setting; the camera makers' own raw processors do, but for example ACR does not (except for the WB).

However, that is again the histogram of the converted image, which is far from the histogram of the raw data.