Musings on Photography

The History of Exposure Decisions

Posted in equipment by Paul Butzi on May 1, 2007


In the comments on my whining post about meters, Chuck Kimmerle wrote:

I can honestly say that I’ve good luck with my Nikon camera meters, but shoot exclusively in manual mode, using the meter reading only as an initial starting point. From that first exposure I look at the histgram, adjust exposure accordingly, and shoot away.

To be honest, I’m not sure why anyone would shoot in auto mode as it takes THE critical technical decision – exposure – and turns it over to a mindless box.

There are two things I wanted to address, here – one minor, one more major.  The first, and minor one is that Chuck is saying that the appropriate approach to using the meter is to let the meter give you a starting point, make an exposure, and then adjust.  Well, that’s what I do, but it doesn’t mean the meter is working well.  If the meter was working well, I’d almost never need to adjust and make a second exposure.  I certainly agree that using the histogram is essential – and that’s my point.  If the meter did what I do with the histogram, we’d have no problem.

Second point – Chuck suggests that the exposure calculation is a critical technical decision.  There was a time when the choice of exposure was a critical decision, because it was an essential decision in the creative process.  But now, that’s no longer the case (except perhaps for people shooting polaroids for prints, or transparency films with the transparencies as the final product).

For a whole host of reasons, early films had a pronounced toe and a severe shoulder.  If you didn’t give the film enough exposure, your shadows ended up compressed, and if you gave the film too much exposure, the highlights ran up onto the shoulder and the highlights were compressed.  For wide range scenes, you were faced with balancing between the two (and adjusting development of the film).  The Zone system was (and is) a way of not only managing how much information gets captured on the film, but managing the way the tones are distributed, because with the graded papers everyone used, making changes in the tonal distribution was fairly hard (especially if you wanted local changes and not global ones)

But as the technology advanced, we got films like Tmax, and Delta – films with H&D curves that were pretty much straight lines.  These films have very abrupt toes, and so they are intolerant of underexposure, but they have no shoulder and can easily capture ten or more stops of range, so as a general rule it was simplest to always provide generous exposure, because this meant that all the information was captured on the film.  And we got variable contrast paper, which let us not only adjust overall contrast more easily and finely, but also let us change contrast locally in the print.  Suddenly, it made more sense to massively simplify the process – give every negative generous exposure (and thus you’d get ALL the information in the negative) and then sort it the decisions about print density, contrast, etc. in the darkroom later.  The decisions about tonal arrangement, shadown contrast, highlight contrast – all these decisions shifted away from being something you decided at the moment of exposure, and became decisions that were made at printing time.  The goal at exposure time was very simple – never, ever underexpose, because the information that fell below the abrupt toe was just lost forever. 

And now, with the shift to digital capture, we’re facing a variation on the “don’t lose any information at exposure time” paradigm.  With digital capture, though, it’s not a matter of generous exposure, it’s a matter of never, ever letting the highlights clip.  That’s because any tonality that falls off the right of the histogram is, essentially, lost forever, just like the information that fell below the toe of film.   At the same time, we’re faced with cameras that capture limited dynamic range, and if important information falls too low on the scale, the information is lost in the noise floor of the sensor, and gone forever, too.  So the general rule is to ‘expose right’; that is, choose the exposure that has the highlights right up against the right hand side of the histogram, but NOT clipping anything.  We don’t worry about tonal distribution if we’re shooting raw, because that’s what we use the ‘curve’ tool in the camera raw convertor (and the curve tool in PS) for.  The goal at exposure time is just like the goal with straight line films like Tmax – get all the data captured, don’t lose anything.  You can always throw data away later, but you can never summon it back once it’s gone. 

(Actually, you can, just as Glendower claimed that  “I can call spirits from the vasty deep.”  Sadly, Hotspur’s reply “Why, so can I, and so can any man.  But do they come when you do call for them?” applies also to summoning back highlights that are clipped away)

So the bottom line is this – with old style limited range film, with long toes and shoulders, printed on graded paper, choosing the exposure was a creative decision, because the choice you made irrevocably fixed the tonal relationships that would end up in the print.  And, as Chuck would point out, we don’t want to delegate a creative decision to a mere mechanism.  We want to reserve these artistic decisions to ourselves, and if we delegate any decisions, delegate the merely mechanical ones.  This follows the IBM maxim – machines work, people think.

But with the advent of modern straight line films (and VC papers) the exposure decision shifted from a creative one to a mere mechanical one – pick the minimum exposure which guaranteed that all the data ended up in the negative.  And with digital capture, the principle has shifted to ‘pick the maximum exposure which doesn’t clip highlights’ but it’s still a mechanical decision.

If a human can make an exposure (or meter the scene), examine the histogramed result, and then mechanically adjust to find the ‘correct’ exposure, then it’s a decision I think we can delegate in most cases.  Or, at least, we could delegate it if the damn meter actually did this.  Sadly, current meters don’t.

But they ought to.

5 Responses

Subscribe to comments with RSS.

  1. Guy Tal said, on May 1, 2007 at 12:31 pm

    To me “critical” implies something that is not easily mitigated. For some reason I don’t consider exposure in these terms. When in doubt – bracket, and the risk is eliminated at practically no cost. Leave the critical decisions for later when you can take the time to work with more precision.

    Another common oversight is that the histogram displayed immediately after the capture is not derived from the raw file but rather from a much smaller JPG which is generated with certain processing criteria applied. The fact that your histogram may look slightly clipped on the small LCD screen doesn’t always mean you’ve lost the highlights. You’ll be amazed how much can be recovered in the raw converter that you didn’t even know was there.


  2. chuck kimmerle said, on May 1, 2007 at 1:36 pm

    I agree with Guy that bracketing can virtually eliminate exposure concerns, but it’s not usually necessary as the wide tonal range of digital files can handle most scenes, esp. if shot RAW.

    I still, though, stand by my position that exposure is THE most important technical decision. Bracketing, while useful, is not always an option as we often get only ONE chance at an ever changing scene.

    I will agree that the latitude of digital files allows a certain degree of error while still producing a acceptable detail, but recovering “lost” details at the far ends of the characteristic curve are not without their costs. In the case of shadows it comes with increased noise and loss of contrast, and in extreme highlights, banding and color shifts.

    I guess the bottom line is to find what works for you and go with it. For you Canon owners, perhaps a change to Nikon? 🙂

  3. Guy Tal said, on May 1, 2007 at 1:50 pm

    I did say “when in doubt” 🙂

    There’s undoubtedly a lot of things Nikon does better – from metering to flash to general ergonomics.

    Now if someone can figure out how to bolt the 5D sensor and imaging chip into an F5 body…


  4. Martin Doonan said, on May 1, 2007 at 2:50 pm

    All this metering stuff got me thinking, which led to a short test. If the histogram (not necessarily metering) is based on a jpg interpretation, do the parameters affect the histogram? this while shooting in RAW.

    Quick check (more work needed) by changing the contrast setting under the parameters, the apparant exposure via the histogram changes by up to 1/2 stop to the right. Needs the contrast pushed to +2. at -2 to 0, no effect. Colour temp, colour space and other parameter settings don’t appear to have any effect. it’s measurable and repeatable.

    I shall be pursuing this further.

  5. […] a couple of posts, Paul Butzi writes here and here about the camera’s exposure meter and its lack of doing the right thing. I guess as […]

Comments are closed.

%d bloggers like this: