09 April 2006

Auto-exposure

Since Phil took the bait, here's a simple test you can do to find out whether automation is homogenizing your photographs. Take one of your photos and open it in Photoshop. Then do this:

Filter > Blur > Average

Image > Mode > Grayscale

Now use the Eyedropper tool to sample any point in the image.

Finally, look in the Color Palette and see what percentage of gray is "average" for your picture.

If the answer is about 50%, you probably used auto-exposure. Phil likes aperture-priority automation. He justifies its use this way:

"I don't think the automation in the camera makes a great deal of difference. Once you have decided what you want to take a picture of, compose the shot and focus on the topic of interest there are only two real choices you can make on a camera; aperture and shutter speed. And the choice of one strongly constrains the other"

This is true - if you want the average density of your photo to be 50% gray. Your camera wants the average density of your photo to be 50% gray, because that's the average for a photo with a "full range of tones" more or less evenly distributed (for example, a picture of a human subject outdoors in the daytime on a grass lawn.)

Here's a photo I took at Reed's late last year. I saw some shadows on the wall in a dark corner and thought they looked interesting in a brooding, film-noirish way. The corner was very dark, and I wanted the picture to look dark, the way my eye saw it. If you run the Photoshop action I've described above on this picture, you'll see that its average gray percentage is 95%.

When I took the picture, I knew that the normal exposure for ambient light in Reed's is about 1/60 of a second at f/2 on 400 speed film. In dark corners, there's much less light. I shot this picture (manually) at 1/30 at f/2. If I'd used aperture-priority automation at f/2 (assuming my camera did that, which it doesn't), the camera would have noticed that the wall wasn't lit, and it would have set a shutter speed of either 1/4 or 1/2 second - resulting in a picture with an average gray percentage of about 50%, which would have looked like this:

I like my picture a lot better than the one an automatic exposure meter would have generated. I could have tricked my F-100's auto-exposure system into producing the picture I wanted by setting "exposure compensation" to tell the F-100 that the scene was supposed to be dark. But then what good is the automation? I already know the scene is dark, and it's no harder to set the exposure values manually than it is to set exposure compensation manually - so all the automation does is make it more likely that I'll get lazy and end up with a bad picture.

Automation won't hurt your "average" pictures (photographs of people in daylight, for example), because those are the pictures it was designed to produce. It is much more likely to hurt dark or light pictures, or anything else "out of the ordinary".

Open up a bunch of your pictures and try the Photoshop experiment I've described above. If your highest gray density is only 10% higher than your lowest, turning off autoexposure will probably improve your photography.

5 Comments:

Blogger Radovan Semancik said...

Good point, but again I cannot totally agree. More here:

http://storm.alert.sk/blog/photography/auto-vs-manual.html

April 10, 2006 4:40 AM  
Blogger Phill H-B said...

Bob, you are still missing my point. The dynamic range I capture in the negative has only a slight relationship to the dynamic range I choose for the print. If I have the information I want captured in the negative I can use post-processing, whether in the dark room or in photoshop to select the effect I am after.

If you look at Ansel Adams prints there is a huge variation in his execution of the same negative.

If the information is captured on the negative it can be removed in the print. If the information is not captured no amount of post-processing can bring it back.

I have not used an F100, However I have used the N90s and F90x that preceded it quite a bit and recently switched to a D50 which shares a lot of F100 features and layout. One of the things I do find problematic with the D50 is that using exposure compensation is much harder than it needs to be and is too easy to forget. The only saving grace is that as it is a digital camera feedback is instant and you can see the histogram instantly.

So one might hazzard a guess that part of what went wrong on the F100 is that a lot of the technology from their digital cameras had transfered to their film range.

April 10, 2006 11:13 PM  
Blogger bob blakley said...

Ah - you're right, Phil, I did miss your point. And it's a good one - but to a certain extent we're talking at cross purposes.

I learned to expose properly from Adams, and he helped me to see properly too. But I don't work the way Adams did. Since I grew up exposing mostly color slides before digital came along, I didn't have a lot of room for post-exposure manipulation. I try to get the exposure I want on the film, which, like most digital sensors, has a very limited dynamic range.

This taught me to make tradeoffs - when the scene exceeds the dynamic range the "sensor" can handle, I need to change something about the light, or sacrifice the shadows or the highlights.

I do make choices at the printing stage, but I also make choices at the capture stage, and, as you correctly note, some of these choices are final, as they cause me to "throw away" scene information which I can't recover later.

What I'm trying to do in this entry is to make people think about the first set of choices, so that they don't need to do too much post-processing.

By the way, the F-100 is very ergonomic - it's the direct successor to the F90, which was also a great camera, and it preceded any Nikon digital SLR (and in fact any DSLR from any manufacturer if I recall correctly). It actually makes it EASY to change the settings I'm complaining about - I just don't want to use them at all....

April 11, 2006 9:13 AM  
Blogger Phill H-B said...

OK Bob, if you want to do slides then OK, but when are you going to show them?

These people are not ready for such knowledge Bob. Inflicting a slide show on the family is even worse than dithering over a photo.

Actually the D1 came out somewhat earlier, I remember looking at it when I bought my N90s. With a price tag of $4500 for a camera that would clearly be obsolete in a couple of years it was not the one to get.

If the F100 had been available at the time I would have bought it.

I can't recall when the D70 and D100 came out, but they are both essentially cut down versions of the D2 with a few features removed. I took a look at the D100 and seriously considered getting one but having used an FG for 20 years getting a second film camera in 3 years did not seem justifiable.

The big difference between the D1 and the D2 is that the D1 was really designed for the news photographer who was willing to sacrifice quality for the convenience of digital. By the time the D2 came out the lower resolution of digital vs 35 mm was pretty much eliminated. Variation in the film stock was a bigger issue than any loss of resolution.

I think that the D50 is probably the most interesting recent camera because it is the first DSLR that takes a better picture than equivalent priced 35mm film. It appears that at some point the 35mm format DSLRs should exceed the resolution of medium format.

April 11, 2006 8:44 PM  
Blogger EdNaz said...

I really enjoy reading the "automation versus manual" debates that go on everywhere. But I think "it depends" is probably the most honest answer to which approach is "right." A couple of photographers whose work I respect (and if you can make what they make as a photographer...well, they probably could care what I think then) have jokingly said that the "P" setting on the Nikon SLRs means "professional" and not "program" mode. Yup, they shoot 90% in P mode. Another digital debate is RAW versus JPG, with armies of photographers swearing that RAW is absolutely superior to JPG, while a few excellent advertising and fashion shooters shoot JPG with no one the wiser.

And then there's matrix metering versus spot versus center weighted and how to best meter a scene, a debate that's gone on for decades. A photog I really admire for his long career of stunning work is Jay Maisel, who might as well glue his camera on matrix metering. He tells a story about getting an excited call from a friend who said he'd figured out the perfect way to calculate exposure. After 5 minutes of details, the explanation ended with "and then I bracket up two stops and down two stops in half stop increments." Hm.

I'm an F100/D2X shooter. I use auto everything in some situations, like shooting events where light is different every three feet. I use manual everything in others where I know exactly how I want to shift an image in final production(oddly I flip to manual instantly when shooting really fast action - I'm still faster at focusing than my camera.) But appropos the Jay Maisel story, when I'm shooting with fine art intent I often capture five images quickly at a range of exposure values, which gives me nearly a 10 stop range of image to work with. It lets me make post-processing every bit as much of a creative process as was seeing and capturing the light. I could (and did) do this shooting transparency film with the F100, an expensive strategy. Still, it paid off more than once by having an exposure that was completely off the wall "wrong" that worked incredibly as a print.

I think that reflects my take on all of the "this or that" debates - "auto" is just another creative tool. Knowing when and how to use it is what sets the experts off from us earnest folks. And my personal opinion only, I think digital now can outperform 35mm in perceived resolution, shooting portraits with a D2X you discover that everyone has these incredibly visible pores on their face, with little hairs in them, that you never noticed shooting 35mm film. Extremely high resolution is not kind to people with normal skin. Having shot Bronica 6x6 with Astia side by side with my D2X and compared large prints - the D2X wins.

June 06, 2006 3:42 PM  

Post a Comment

<< Home