High ISO or Boost the Exposure in PP

[SwissJon]

OK, As first discussed in this post, I've been doing some tests to see which is better, High ISO v's boosting the Exposure of an under exposed shot in Post with some interesting results. I've tried my D7000 in RAW, an Olympus XZ-1 in RAWand a Panasonic TZ-10 in JPEG. (The last two are P&S's.

This is the first time I've seen these cameras alongside each other. It's very interesting for me, because aside from anything, it really shows me how much better the D7000 is against the compacts, and that RAW vs JPEG doesn't really seem to acheive much in this test, and that the compacts IQ is basically crap at anything other than base ISO.

OK, to allow others to post their test results (if anyone is interested in joining in this little experiment, I'm going to post 2 photos, one correctly exposed at ISO 1600 and one with exactly the same speed and aperture settings at ISO 400. This will make the ISO 400 picture 2 stops under exposed. I then changed the exposure setting in Lightroom by +2.00 and then exported the JPEG files at the highest quality possble and then cropped the centre to 640x480. No other alterations were made to the files.

The files are found in the following posts, my comments below each pair. In case you don't recognse the subject, it's a Sky tv remote control. Seems like a good subject. It has colour. The background is about 15% grey and the buttons are high contrast.

[SwissJon]

This is a Nikon D7000 with an 85mm 1.4 prime lens. F/1.4 1/400s Cropped from a 16Mpix file.

ISO 1600



ISO 400



The thing I noticed here is that the ISO 400 seems to have more colour saturation. The level of noise is pretty much the same. Seems to me it's slightly better shooting under exposed than shooting at high ISO.

[SwissJon]

OK, so here's the results for the olympus shooting in RAW. f/2.5 1/400s cropped from a 10Mpix photo.

ISO1600



ISO 400



OK, so the quality here is much much worse. The two images are almost identical.. I wouldn't say either is usable.

[SwissJon]

OK, the last of my test Panasonic TZ-10 f6.3 1/25s cropped from a 12Mpix photo

ISO1600



ISO400



In this case, I'd say that the ISO400 under exposed shot produced a marginally sharper and more saturated picture.. Other than that little difference.

[Rentham]

I did a similar experiment a while ago. The images are gone from my flickr (I'll try to find them again if anyone is really that curious) stream, but the general take away for me was underexposing and pushing in post produces slightly better results.

[OsmosisStudios]

Now try this: expose properly.

Your images are almost all under exposed; only the ones from the panasonic are "properly" exposed. You'll see a difference then, I can assure you.

That said: I actually disagree with your findings. For the D7000, the blacks get cut too quickly when pushing from ISO400; you're not getting the same dynamic range, and you'll notice it even more with proper exposure. Not to mention: shooting wide open isn't ideal; stop down to f/2.8 and you'll see a difference there alone. On the Olympus files, you're getting a similar issue: your blacks are being clipped in the ISO400 image.

[andrewdt]

The thing I noticed immediately on the D7000 shots is that on the ISO400 shot the color saturation made the "abc" appear slightly more fuzzy around the edges which might cause us to sharpen more? I would also guess that this low light example is as good a test for this excersize as possible simply because in good lighting we would not need to even try it.

[sk66]

I've advocated underexposing and pushing in post for a long time...but it's not that simple.
You have to know when you reach the signal noise floor for your camera...ISO's above that level will generate excessive noise for the exposure as compared to post work...also depending upon your RAW processor/flow. Up to the signal floor you are better off pushing ISO (sacrificing dynamic range).

I've had a D7000 for a while and my *impression* is the ISO performance is *similar* to my beloved D3...
That means the *floor* is around 1600-3200. I *think* with a D7000 if you took an ISO 1600 image 2 stops underexposed, pushed it 2 stops in post and then compared it to a "properly" exposed 6400 exposure the 1600 would be "better".... But you can't just arbitrarily say pushing in post works better...it doesn't always. (nor can you say using higher ISO is always better)

[SwissJon]

Originally Posted by OsmosisStudios

Now try this: expose properly.

Your images are almost all under exposed; only the ones from the panasonic are "properly" exposed. You'll see a difference then, I can assure you.

That said: I actually disagree with your findings. For the D7000, the blacks get cut too quickly when pushing from ISO400; you're not getting the same dynamic range, and you'll notice it even more with proper exposure. Not to mention: shooting wide open isn't ideal; stop down to f/2.8 and you'll see a difference there alone. On the Olympus files, you're getting a similar issue: your blacks are being clipped in the ISO400 image.

Interesting you should say that because the cameras meter disagrees. When I took the photo I made very sure that the meter reading was dead centre when I set the speed and aperture for ISO 1600, and no surprises, the 400 was 2stops under. The histograms agree, they do bunch to the left but that, I assumed, is because there's very little white in the photo, the brightest highlight being roughly15% grey. I think the "properly exposed" photos were due to the differences in metering methods. Spot metering on the two "underexposed" examples, and larger centre weighted metering on the other, which would have encompassed the black buttons and metered a brighter picture accordingly. The photos are therefore exposed exactly as you'd expect, putting the 15% grey at mid tone. In fact, when you look at the actual subject, the Panasonic photos are overexposed compared to real life, making the mid tones into highlights. The other two are a lot closer to what I was actually seeing.

My intention with this photo was to do exactly what I did. Shoot to the meter, rather than shoot with my eyes. I didn't want to interpret what I was doing as a photographer, I wanted simply to collect empirical data. What's interesting is that the two cameras that shot RAW have clipped blacks compared to the JPeg. One would have expected this to be the reverse.

[Biomech]

I'm not too sure my camera would produce similar results, in terms of noise. I certainly plan to try it out tomorrow, I'll post the results