Correct exposure is everything

[Alis]

I have recently (too late?) come to the conclusion that if you don't get the exposure right, in the camera, at the you capture the image, there will be no way you can get a good image out of that file, no matter how much post-processing you apply.

I mean theoretically and to a large extent practically. What do you guys think?

Alis

[Dave Humphries]

Hi Alis,

You're probably right and I am mis-guided, but this seemed to come up alright.

When I was processing in ACR, it brought it up 1.5 stops, as you can see from original vs processed images in that post.

But then I am not trying to get subtle nuances on skin tones as you often are - and the eye+brain is especially good at differentiating skin tones, where as one black and white bird looks much like another!

Regards,

[GillR]

Getting the exposure spot-on in-camera is always the aim but, in cases of doubt, I would always prefer to have a slightly over-exposed/exposed to the right shot - as long as the highlights aren't blown of course.

This blog post, and the one it links to, show what I mean.

But that being said, I don't photograph people (ouside of family snapshots) so I have no idea how well, if at all, that would work for skin tones as opposed to landscapes or macro!

[Ashwin]

I have recently (too late?) come to the conclusion that if you don't get the exposure right, in the camera, at the you capture the image, there will be no way you can get a good image out of that file, no matter how much post-processing you apply.



Alis

Hi ali
nothing like GET EXPOSURE RIGHT at the time of capturing Image,
It is difficult if not impossible to get perfect image every time.. If need be, you can
improve your image by Post-processing

[Colin Southern]

I have recently (too late?) come to the conclusion that if you don't get the exposure right, in the camera, at the you capture the image, there will be no way you can get a good image out of that file, no matter how much post-processing you apply.

I mean theoretically and to a large extent practically. What do you guys think?

Alis

Hi Ali,

I think that a lot depends on the dynamic range of the scene, and whether or not your shooting RAW.

... If your shooting JPEG then - yes - you have to get it pretty close "in camera" (although you can push them 1 to 2 stops without too much noise becoming apparent).

... If your shooting a scene where the dynamic range pushes the sensor to the limit then - yes - you have to get your highlights pretty much up to the limit (RAW is a given for this type of shot).

... If your shooting RAW and are shooting a relatively flat scene (eg no specular reflections or back lighting) then you'll probably have around 12 stops of dynamic range available to capture a scene that's probably only going to require 5 or 6, so you'll probably get away with a 3 or 4 stop under-exposure to a large degree (although not recommended!).

Over-exposure with digital is fatal (barring a little leeway that can be extracted from a RAW file). When shooting landscapes - if you can - you'll often get a nicer result if you under-expose a stop or so - more saturation and highlight detail, although I'm still trying to fully understand why this is - possibly part of the problem is that we photographers are thinking in terms of a single exposure whereas the camera is really dealing with 3 exposures (one for each channel); if the colour temperature is relatively neutral then it's not a problem, but when one channel leads the other by a big amount (eg red leading blue with the warmth of a sunset) then getting a "good" capture (as opposed to a "technically correct - shoot to the right type exposure") is definately a whole new ballgame.

[Bill44]

Colin's suggestion for underexposing is like the old transparency trick. Underexpose by one stop then shove more light through it. Quite often a far better result.

[will_c]

If you're recording raw files then I think that you can make quite significant adjustments to exposure after the fact. This is closely related to the ability to white-balance out a strong colour cast.

The attached pictures (from the famous 'violin on a couch' series) were taken in a room in which the only light source was a 60W tungsten bulb under a yellowish-cream coloured shade on the other side of the room. Even using the tungsten preset the scene still looks distinctly yellow, as does the white card that is used as a colour reference. The more normal-looking picture has been white balanced using the card. I also added +0.5EV to the overall exposure for the two versions of the scene during processing. These were all taken at ISO 800 with a 40D.

The RGB histogram of the white card has the red 1 stop down (from the top), the green 2 stops down, and the blue 3 stops down. The luminance histogram is 2 stops down, where one would expect it to be for a normal exposure. So to follow on from Colin's post, if the light, or part of the scene, has a strong colour cast to it then one might well need to allow some extra headroom in order to avoid individual channels clipping, even with a nominally correct exposure.

Personally I am quite impressed by the ability that the 40D (in my case) provides to dig a reasonable result out of quite difficult light. Similarly there is quite a bit of leeway to adjust overall exposure (provided that it's not overexposed to begin with of course). You have to have a raw image to work with though. JPEGs, which are already gamma compressed for a particular exposure, provide much less latitude - large adjustments will quickly lead to colour-banding effects and so on.

Will

[Alis]

Thanks everyone!

OK, if I understand correctly, at least practically, there is no degradation of the image quality (colors, saturation, etc) if the pixels are exposed inappropriately (within some reasonable rescuable range). You can always compensate for the intensity of light and the right colors will come back automatically. Is this correct?

Then I don't understand what those guys are doing all the time in the studios (a controlled environment, nothing basically changes, subject is always sitting in the same place, lights are always at the same location) with a meter, measuring exposure all over the place? At worse, they can just bracket the shots and later decide which one looks better. Right?

[Bill44]

In a studio situation they quite often have lights/flash that needs to be adjusted manually. Also, if you look at some model shots, there is quite a variety of ways to light a subject. From experience a good studio shooter can judge how much light they need to direct at a certain spot, and they use the meter to measure the existing light because the Mk1 eyeball will adjust itself to the available light and is therefore unreliable.

Even using modelling lights to get the right mix they still need the meter to read the different values.

[Colin Southern]

OK, if I understand correctly, at least practically, there is no degradation of the image quality (colors, saturation, etc) if the pixels are exposed inappropriately (within some reasonable rescuable range). You can always compensate for the intensity of light and the right colors will come back automatically. Is this correct?

In practical terms - for all intents and purposes - yes.

Then I don't understand what those guys are doing all the time in the studios (a controlled environment, nothing basically changes, subject is always sitting in the same place, lights are always at the same location) with a meter, measuring exposure all over the place? At worse, they can just bracket the shots and later decide which one looks better. Right?

It's not just the intensity of the light with regards to exposure in a studio situation; to what degree the light varies from spot to spot is also a consideration (delta EV), and that's also what's being measured. Keeping in mind too that if your setup for one type of shot - with one person - in one position then yes, very easy to get things right, but in a studio your normally shooting different types of "looks" and different numbers of people resulting in constant lighting changes, hence the need for re-metering. Bracketing is one possible solution, albeit a fairly poor one - it only brackets in terms of overall exposure, not in terms of - for example - the amount of light reaching the subject from a key light -v- that on the other side of their face from a fill light. Far easier just to get it right in the first place! (keeping in mind too that studio strobes are usually NOT ETTL integrated - it's all manual baby!

[Colin Southern]

In a studio situation they quite often have lights/flash that needs to be adjusted manually. Also, if you look at some model shots, there is quite a variety of ways to light a subject. From experience a good studio shooter can judge how much light they need to direct at a certain spot, and they use the meter to measure the existing light because the Mk1 eyeball will adjust itself to the available light and is therefore unreliable.

Even using modelling lights to get the right mix they still need the meter to read the different values.

Why didn't I write my ramblings as neatly as this!

[Alis]

OK, now I am confused. Are you guys for spot on exposure or the sort of "anything-goes-fixed-it-in-postprocessing" type of exposure?

After all, the studio people can selectively fix all those different areas of exposure in postprocessing. Can't they?

[Colin Southern]

OK, now I am confused. Are you guys for spot on exposure or the sort of "anything-goes-fixed-it-in-postprocessing" type of exposure?

In a studio setting you'll be mainly dealing with reflective objects, and with a low dynamic range - easily handled if shooting in RAW with any modern SLR camera, so "ideal exposure" isn't "life and death", but having said that, why rely on your safety margin to save the day when you have the option to get it right "in camera" in the first place?

After all, the studio people can selectively fix all those different areas of exposure in postprocessing. Can't they?

Yes - but - at the (literal) expense of a lot of time and effort. Perhaps it's a bit like saying "why bother with diet and exercise when you can just book in for a quadruple bypass later on in life?"

[Bill44]

OK, now I am confused. Are you guys for spot on exposure or the sort of "anything-goes-fixed-it-in-postprocessing" type of exposure?

After all, the studio people can selectively fix all those different areas of exposure in postprocessing. Can't they?

Personally I like to get it quite a bit better than "anything goes" and that is obtained, in general shooting, by the camera metering and experience regarding any compensation required. In sports action where you don't have much/any time to play around, I'm quite prepared, if needed, to blow the background so long as I capture the action. In a case like this I will spot or centre weighted meter and to hell with the rest.

As regards the studio shooter, time is money. If you realised the money a good studio shooter charges for a session you would understand why they would rather spend 5 minutes metering than 30 minutes post processing. If they don't get the shot right it can mean a reshoot at their cost, with a consequent loss of income and a damaged reputation.

[Alis]

OK, it is clear now. I have heard the time-saving argument before but thought there should be more into it.

Thanks, everyone.

Alis

[rpcrowe]

With the price of relatively large CF cards fairly low, why not bracket exposure? It is extremely easy to accomplish, especially with my Canon DSLR cameras (I don't know about other brands).

Just select exposure compensation through the menu and select how many f/stops you desire the exposure compensation to be (I will often vary my exposure depending on what type of subject and whether I am shooting at night or in the day). Then select in what sequence you desire the compensation; (I like: normal exposure, then minus and then plus).

If you have selected the burst mode, the camera will fire off 3-shots (in sequence that you selected) and then stop.

Nothing could be easier. I use exposure compensation very often, especially when shooting once in a lifetime subjects. I will definitely use this method on a forthcoming trip to China. I have 24-Gigs of CF memory and a notebook computer (with accessory remote HD as back-up). This allows me about 600 bracketed trios of shots until I need to download. That is usually enough for my day's shooting. And, after-all, 9,000 or so images from a 15 day trip should be enough.

Another advantage is that when using a tripod or some other solid support, I can combine each trio of images into an HDR image if I so desire.

I don't use exposure compensation when shooting sports or other subjects when I want to utilize the burst mode for multiple exposures.

[eNo]

I have recently (too late?) come to the conclusion that if you don't get the exposure right, in the camera, at the you capture the image, there will be no way you can get a good image out of that file, no matter how much post-processing you apply.

I mean theoretically and to a large extent practically. What do you guys think?

It depends: what do you mean by "correct exposure"? If you mean how much light hits your sensor to produce the "correct historgram" you are tredding on thin ice. But even if we go with a well-contained, non-clipping histogram as the golden rule, how correct do you have to be? In my experience, 0.7EV either way is indeed fully addressable in PP with little if any negative effects on the final image. It all depends on the scene and its lighting characteristics. In addition, in some cases you need to under-expose, as for instance when you want richer, more saturated reds. Thirdly, in some cases over-exposure or clipping of certain image areas or colors is unavoidable and even desirable -- example: street lamps will clip in nighttime scenes.

Above all, however, "correct" is only meaningful in light of what you want or need out of the image, IOW your purpose for the image. You really can't separate exposure, namely how much light falls on your sensor, completely from compositional considerations. If your subject is in shadows and the background is in bright noon sunlight, the "correct" exposure (unless you put your camera away and come back later) will be a well exposed subject and terribly over-exposed background. Some would look at the overall exposure and say it was a failure, but it's up to you to make the final judgement because only you determine the purpose of the image. For my fuller take on this, check out this article I wrote a few days back:

The quest for the correct exposure

[Colin Southern]

With the price of relatively large CF cards fairly low, why not bracket exposure?

I quess that the flip-side question (especially if shooting RAW) is "If your confident with your exposures, then why bother?" (Expecially with exposure aids like highlight alert and the histogram).

[William W]

Over-exposure with digital is fatal (barring a little leeway that can be extracted from a RAW file).

The word "fatal" is a key word to this thread, IMO.

There is so much talk about “Expose to the Right with Digital” that I perceive (and have seen the irrecoverable results) when this technique has become FATAL for many.

I work a lot with white textures (Wedding Gowns). Blowing the Whites in the Bride's Gown is one of the TWO most common FATAL errors displayed by those taking on the task of Wedding Photography - and it is also common with those who think they are Wedding Photographers with credential also. (FWIW the other main problem is not understanding / respecting DoF).

It seems to me the "Mantra" ** of "Expose to the Right" is often interpreted as "Overexpose, Shoot RAW, Recover Later".

(**I like that word Colin - you used it the other day elsewhere )

As a result of a debate on this particular topic on a Pro Wedding Forum - I am just embarking on a series of tests with a colleague in San Francisco - we use the same gear – we wish to establish exactly what the limits are, shooting RAW.

We hope these tests will also establish a more precise detail of the general "F16 Rule" - for maintaining Detail of White in Sunlight and Sunlight + Flash Fill.

Also I hope to establish if there is a seasonal / hemisphere / ozone / smog variation which accounts for any differences in our two exposure / meter readings at the same time of day in Bright Sunlight. (S.F. and Sydney are basically the same latitude)

We both, just from anecdotal experience, are of the opinion that the "latitude" of RAW is about 1/3 to 1/2 of a stop over, before the detail is unrecoverable – that’s not much before “Fatal”

WW

[carregwen]

The camera can only do so much with exposure, especially in more extreme exposures. Even the human eye can't cope all that well. Imagine driving down a high-banked lane on a dark road surface. Suddenly the sun comes out from behind a cloud. Your eyes will struggle to balance the competing light levels, and what do you do? You reach for sun-glasses, or pull down the visor.

This is a shot with some blown detail in the BL corner. This is as it came out of the camera in RAW



In CS4 I used the grad tool to apply an under-exposed grad from BL corner towards the centre. It is still unbalanced in the rest of the shot though.



In PS proper, I edited the TIFF. I selected the grad area I had applied in ACR and feathered and inverted that selection. I could then work on the rest of the image using Levels etc to bring back some more exposure. I think that's the best you can do.