Why did exposing for the highlights and raising shadows in Adobe Camera Raw result in so much noise?

Question

I'm wondering why this picture is so grainy. It was shot at f16, 1/125 seconds, with an ISO of 100. I underexposed it to maintain the highlights in the sky but that is what I always do, and it doesn't end up looking like this! I edited it as I normally do in camera raw/photoshop, but I didn't take any other measures in post because I just wasn't sure what to do about it there, either. It seems unsalvageable. The other ones in the batch seem to be showing signs of grain, also. So I would like input on how to

1. prevent this from happening
2. ameliorate the situation in post, if it does.

Thank you! @null, I managed to add another photo. It is grainy--sorry, noisy!--also, but not as bad. It was shot at 1/160 seconds, at f/16 and ISO 100. Looking at the histogram I now see that they really are too underexposed--sorry to have bothered everyone with such an easy question! Never had so much noise before in landscape and I sort of panicked rather than putting on my thinking cap!

Answer 1

To clarify, here is what you are apparently talking about:

This is a 1:1 crop from the middle of the picture.

There is no grain here. What I see looks like sensor noise. "Grain" makes no sense for such a digital picture anyway. Note that the clouds in the background look fine, but there is significant noise in the much darker foreground areas.

Basically, this scene has too much dynamic range for your sensor. To fix this, either take multiple images at a sequence of f-stops and stitch them together with a technique called HDR (high dynamic range), or get a better sensor. This is one area where sensor quality really matters. It would help to know what camera this was taken with. If this was a cheap point and shoot, then a better camera would probably make a significant difference. If this was a Nikon D3s, then your only option is multiple exposure HDR. (However, having experience with a D3s, I think it could do this natively with significantly less noise that exhibited here.)

Of course this is all assuming you exposed optimally for the sensor you do have. That means the brightest areas (the whitest parts of the clouds) need to be almost at the brightest raw values. Using the 0-1 scale for raw values, if the brightest areas were .9 or so, then you've done what you can. On the other hand, if they were .5, then you wasted a whole f-stop of precious sensor dynamic range. 0.25 would be 2 f-stops, etc.

You say you underexposed, but relative to what exactly? Some cameras have a mode that does exactly what you want in this case, which is to expose the highlights to hit just under the top of the sensor range. If that's what your camera was already doing, then underexposing only wasted dynamic range.

Again, it would be good to know the camera and what the histogram of the raw data actually was.

Answer 2

I underexposed it to maintain the highlights in the sky but that is what I always do

But the highlights aren't always the same and the dynamic range of two scenes can be different.

Exposing for the highlights is exactly that: it exposes in a way that the highlights aren't blown out. It does not guarantee anything about the rest of the tones.

Highlights aren't everything. It's merely one end of the histogram.

Let me try this analogy: If you are a tall person and you walk into a house with low ceiling then you might bump your head somewhere. You could be inclined to say "How is this possible? I'm standing on the ground with my feet, how could I possibly bump my head?" And the answer is that standing on the ground does not guarantee sufficient ceiling height.

To prevent this from happening, you should capture enough of the exposure range. this means either using a camera with more dynamic range or bracketing the exposure. (taking many exposures, each one exposing one part of the image correctly), then merging them in post processing either via HDR process or by masking each exposure so that only its "good" parts are visible.

To fix it in post if you only have that single exposure, you'd have to apply more noise reduction I'm afraid. I agree that it looks unsalvageable. The amount of noise is massive. With a lot of noise reduction the details will suffer. You won't be able to print this image large, but maybe as a small thumbnail somewhere it might be useable.

Answer 3

Bracket. Espose the sky and the ground as separate exposures, and then combine them. What you're trying to do has become known as HDR and there are tools (even buikt into camera firmware!) Specific to that.

At your preferred ISO 100, you ought to learn how far down in the shadows you get before the noise shows. That is a limit to how much you can raise the darks. In lesser uses, if you raise everything by (so much) you expect noise to start appearing (where in shadows).

Just because it's innthe histogram doesn't mean it's all the same quality.

Answer 4

The less light that is captured by the sensor, the more likely noise is to appear. (See: this dpreview article). Basically, think of the pixels on the sensor as buckets to capture photons. The more photons that drop into the bucket, the better the signal. But the fewer photons, the less signal you get, and the higher the proportion of the data that you get is noise. Upshot: random noise is always more prominent on the darker areas of an image (fewer photons) than in the lighter areas (more photons).

When you digitally "push process" (i.e., underexpose and then bring up the exposure in post), you'll always increase the noise. This is why a lot of folks use a technique called "expose-to-the-right" (or ETTR)--where they adjust the exposure to be as high as possible without blowing any highlights, and then do digital "push processing" (i.e., "overexposing" and then bringing down the exposure in post) to reduce noise. Another technique, Guillermo Luijk's ZeroNoise algorithm, uses a +4EV 'overexposed' shot to eliminate noise.

Generally speaking, the way you underexposed in camera and then pushed in post is the easiest way to bring out the noise in an image. To avoid it, the easiest things to try would have been ETTR, or bracketing exposure over a series of images, and then combining different parts of the those images together to get "properly" exposed with less noise. There are dozens of techniques for this, such as HDR processing, exposure fusion, masking, or (if you shoot specific Canon models) Magic Lantern's dual-ISO mode.