Why is auto exposure so useless?

[SZRitter]

No light meter in the world can know your mind.  You can help it along with a little preference compensation, but all you do is get close.  The extended DR of newer cameras is certainly helpful, but if post processing of RAW is involved, then you might want a totally different exposure for the file versus the finished image.

Which, if I'm not mistaken, is part of the point of the Zone System. All meters try to get Zone V (i.e. middle gray), and if I use think most of the scene is actually in Zone IV, just add a stop of compensation. The meter didn't "lie", it did what it was supposed to. It's just your job as the photographer to figure out how to interpret the results.

[Slobodan Blagojevic]

... Move that spot meter around and you get different readings.  Which is "CORRECT". [?]...

Every one of them.

[mouse]

"Why is auto exposure so USELESS?"

Such blatant hyperbole is a common trick with those who enjoy trolling.
I am surprised it succeeded in catching so many fish in this forum.

[Colorado David]

As this topic has aged, I've typed up a follow-up response a few times and then not posted it.  I'm convinced that when people make these types of declarations; Why is auto exposure so useless?, they're not really interested in a discussion.  They don't want others to disagree with their post.  They only want a bunch of followers to say how brilliant they are and how much they agree.  I think they're disappointed when an actual discussion takes place.

[Rob C]

Incident or spot not really the issue.  The problem is expecting their to be a "TRUE" reading.  Move that spot meter around and you get different readings.  Which is "CORRECT".  Incident solves this to some extent by taking the reflectance difference of surfaces out of the equation, but try in dappled light!  

That's not the way to use a spot meter nor any other reflecting light meter. You should point it specifically at a mid-grey, which can be any colour, as long as it's a mid-tone. The difficulty isn't for the meter, the difficulty is for the photographer to know what a mid-tone really looks like, which is why they sell grey cards.

In my own battles with reflected light metering, I used to measure the darkest area I had to have with detail, and expose for that, which would sound like over-exposure, but I always cut development vis-â-vis the maker's recommendations. You got flatter negs with a longer range.

I wouldn't use that technique for digital: I'd use incident readings, in which case I'd take the reading from the subject position pointing the meter at the camera. That would expose so that it didn't blow the max. white. In other words, trannies and digital work much the same way: respect the highlights.

All the above, with digital, doesn't make a lot of sense if you can work your Matrix meter. The point of introducing the idea of an incident light hand-meter into this thread was to enable the snapper to set his camera (digital) at a single, given combination for all of his shots, just as long as the subject always got the same lighting on it. That way, he could point the camera up or down or however, ignore the message of panic signalled inside the viewfinder, and just keep shooting, because the subject would be exposed properly, and only the sky might get blown, which wouldn't much matter.

Rob

[Rob C]

As this topic has aged, I've typed up a follow-up response a few times and then not posted it.  I'm convinced that when people make these types of declarations; Why is auto exposure so useless?, they're not really interested in a discussion.  They don't want others to disagree with their post.  They only want a bunch of followers to say how brilliant they are and how much they agree.  I think they're disappointed when an actual discussion takes place.

But fortunately, regardless of the OP intentions, which I guess remain private to him, discussing these things can make folks take a deeper look at the why of some technique... not all bad, in the end.

Actually, it's quite interesting to have to reconsider the things one used to take for granted and do without a second thought in a previous life. That's one drawback of digital: it removes curiosity and understanding because so much works by itself, without our really thinking why...

;-)

Rob

[jjj]

That's not the way to use a spot meter nor any other reflecting light meter. You should point it specifically at a mid-grey, which can be any colour, as long as it's a mid-tone. The difficulty isn't for the meter, the difficulty is for the photographer to know what a mid-tone really looks like, which is why they sell grey cards.

Which is very simple really, not sure why other people think metering or manual exposure is so difficult. I used to point camera at my palm which I always had with me unlike a grey card, add 2/3rds of a stop and slides would come out spot on.
You can also learn to calculate the exposure by eye.

[fdisilvestro]

Some thoughts

All metering techniques and devices have their shortcomings. A couple of examples:
- Hand held meters do not take in consideration lens transmission (T numbers) and the effect of magnification at close ranges / macro
- Incident metering is not always an option since you might not get access to where you are pointing your camera and the lighting conditions may be different than those where you are (think about inside a theatre)

Nothing difficult with metering or manual exposure, the only issue is that it might be slow for action, news

You can actually use a spot meter to point at different areas of the scene to understand the actual dynamic range and see if it will be possible to capture it with your camera.

Add flash / fill flash and it gets a little bit more complicated

What it lighting changes during actual exposure? Even more difficult

Auto is not that bad actually, if you understand under what conditions it works or fails (in my modest experience)

Bracketing is here to stay

"He would point it (exposure meter) in several directions, take a reading from each, and fiddle with the dial with a thoughtful expression. "It says one-quarter second at f/32, I'll give one second"

Ansel Adams, The negative, about how Edward Weston determined exposure

[Justinr]

Let's see if I understand this correctly.   We can't trust the camera's auto exposure ability.  I have heard it is "USELESS".   So we'll shoot manually.   That's the solution!!!      But how do we know if our manual exposure is correct while we're shooting? 

Oh wait, of course, I've got it!   We'll check the histogram to be sure our manual exposure is correct.  I guess it must be OK to trust the camera's determination of what the histogram should look like.  Right??  Or maybe the camera's histogram can't be trusted as we've seen discussed here many times before ("We want a RAW histogram, not some crummy jpg rendered histogram that isn't accurate!!").

Hmmm.  OK, NOW I've got it!  We'll review the captured manually exposed image on the camera's LCD to be sure the exposure is correct.  It must be OK to trust the camera's jpg rendering on that little 3" screen.  Right??  or maybe we can't trust either the camera's jpg conversion from the RAW data or the LCD's ability to faithfully show the camera's converted jpg on the LCD !!   

After all, if we can't trust the camera's auto exposure capability to generate a file we can use, why should we trust anything else that the camera does that might help us get to an exposure that we decide is correct?

Further variations on this theme:
Why trust the incident meter because......
Why trust either the camera's or my hand held spot meter because......
Why trust that my monitor's rendering of the image file because........

Brad

One tiny point, A histogram is a representation of what has been recorded, it is still the operator making the decision as to whether the exposure is acceptable, not an algorithm that references 30,000 representative images and cooks up the camera settings based on what a programmer thinks ideal from his Asian office.

Fom my point of view, trying to take multiple shots in a rapidly changing environments, AE has no use.

[Justinr]

Justin,

I found a solutions for you:

Alas, I have no iphone. But listening to my families constant grumbles about lousy battery life and poor reception on their various smartphones perhaps that should be a 'hooray' as my ancient old mobile will go a week without recharging and I don't have to stand on the loo with my head out of the window to hear callers.

[Justinr]

I believed you the first time. Are AE "poor exposures" so unlikely that you won't be able to make some more tomorrow?

Well I was out on another job today and I could have spent time trying various exposure modes etc but to be honest neither I nor the farmer were prepared to spend all day at it. Out of the 116 shots I took only four were too far out of the acceptable range, ie, I wouldn't be able to recover them successfully,  and two of those were sighting shots that I didn't delete in the field. All done manually with varying levels of light as the sun played peek a boo between the clouds.

Does anyone get a 98% success rate with AE?

The attached is just about spot on for this job although I'll need to freshen it up and get rid of the telephone line before sending off.

[pegelli]

Does anyone get a 98% success rate with AE?

It depends what you call "success", but here's some data from participating in a FotoJam in Antwerp last Sunday. Shooting conditions were varied, harsh sunshine, shadow scenes, inside a tunnel, the station and various churches. Light was all over the place.
I took about 250 shots with two cameras and exclusively used AE (matrix metering mode). I did frequently use the + and - EV option based on the situation I was faced with (backlight/frontlight/contrast).
Of the 250 shots I left 167 shots in my Lightroom library for the moment, the deletes were never because of exposure failure, but mainly doubles, undesirable composition, missed focus.
Of these 167:
- 9 shots required +/- 0.2 EV in the raw converter
- 8 shots required between 0.3 and 0.5 EV in the raw converter (positive and negative)
- 2 shots required between 0.6 and 1.0 EV in the raw converter (positive and negative)
- 3 shots required + 1.2 EV in the raw converter (because I left my -EV compensation at -0.7 in the camera by mistake while they actually needed +0.3, it was the end of the day and I was tired)
All other shots do not need any EV correction in the raw converter
All shots are easily tune-able to a desired end-state in my raw converter, didn't spend more than 2-3 minutes per picture on average.

All-in-all I'm happy how the camera's AE system performed during the day.

[Justinr]

Until you show some example of the AE failures you claim happened, there's not a lot more to be said.

Quite so, so lets not say anything. I got the shots I wanted, manually, with two different cameras and with no time spent fapping about with AE modes or endlessly arguing over the finer points of camera workings.  I now have the article to write based on the conversation I was holding with the owner while shooting it.

Pegelli got his shots with AE and so he's happy, as am I, so that's two happy bunnies! Why spoil it? 

[pegelli]

Pegelli got his shots with AE and so he's happy, as am I, so that's two happy bunnies! Why spoil it? 

Maybe rename the thread to "why I don't like AE?" and then indeed just lock it, no point in stringing it out further.

[Justinr]

If you have no intention of showing some example of the AE failures you claim happened, you might as well lock the thread, because some of the criticism voiced in this thread starts to seem more valid.

No, I won't be locking the thread because I am more than happy for folk to have their say whether it be in my favour or not, it's a sort of free world and I would only encourage it to remain so.

As an old hand of fora I know damn well that characters will get on their high horses to spout off about the subject mainly in an effort to demonstrate some sort of superiority and I really haven't the time, energy or patience for that sort of malarky nowadays.

The situation is that I find AE unsatisfactory in the field and I know that other professional users do to. Lads on here can shout and stomp and sneer as much as they like while using every device to demonstrate that we are all idiots and know nuffink, but may I point out that those who actually get commissioned for this sort of work do so because they produce decent images, three of us met in the field last month and were somewhat relieved to find that we all share the same opinion of AE.

But carry on anyway.

[pegelli]

The situation is that I find AE unsatisfactory in the field and I know that other professional users do to. Lads on here can shout and stomp and sneer as much as they like while using every device to demonstrate that we are all idiots and know nuffink, but may I point out that those who actually get commissioned for this sort of work do so because they produce decent images, three of us met in the field last month and were somewhat relieved to find that we all share the same opinion of AE.

And this is not being shouted from a high horse?  

[Justinr]

And this is not being shouted from a high horse?  

Sorry for taking your toy away.

What's a Fotojam BTW?

[pegelli]

Sorry for taking your toy away.

You didn't, I never throw my my toys out of the pram (nor claim superiority based on getting paid or any other reason like you do)

What's a Fotojam BTW?

Google is your friend (and so am I, you just need mature a little more to realize it) FotoJam

[Justinr]

You didn't, I never throw my my toys out of the pram (nor claim superiority based on getting paid or any other reason like you do) Google is your friend (and so am I, you just need mature a little more to realize it) FotoJam

What's your problem, do you want to share it with us?

[dwswager]

This real topic to be discussed is "For which shooting situations and under what lighting conditions is it advantageous to use AE versus Manual and how to get the most predictable results from AE?"

Metering your palm is fine if it is in the same light as your subject.  Otherwise it is detrimental.

You always have to calibrate the meter you are using.  Otherwise it will be consistently wrong.

My AE is 100% accurate...that is 100% of the time it sets the same expsosure in the camera that it reads when it metered!  Doh!

Oh, with film we were attempting to get the best exposure because the limited DR for film, especially slide film.  With Digital now it isn't quite that limited.  My D810 allows not only more forgiveness, but more flexibility.  Now we are trying to maximize quality and minimize post processing.

Finally, a histogram is usually a converted representation based on a specific conversion profile. I have found custom calibration profiles to require significantly less exposure adjustment in post.