Why is exposure correctable in RAW [duplicate]

Question

I have been reading some tutorials and documentation on Rawtherapee software, specifically its exposure compensation.

I am curious how exposure is correctable after the RAW image is captured?

For example, if I take a photo of clouds that is overexposed and there is a white clipped area, how can rawtherepee correct that clipped area if all it know is that there is a bunch of white pixels. How can it restore the white pixels to non-white?

Is exposure just a post process value that is contain in the raw file, but not a characteristic of the actual raw sensor data? If this is the case, why even apply exposure in a RAW image and not just always after the fact?

If my question seems unclear it is probably because my conceptions are unclear to begin with.

Answer 1

Very good question :0)

When you expose your image you have more detail than you can normally see on a normal monitor. This is more a limitation of the display technology than the raw sensor.

This technology is limited to the 256 levels of gray for each channel, but your sensor captures more levels on each channel, this means you have more color diferences on the dark areas and bright areas, making it possible to "reinterpret" your image as is it was exposed brighter or darker when you see it again as 256 level image.

I think a key word here is reinterpret. Remember that a raw file is considered an "undeveloped" image.

Remember the old days when you used negatives. You could expose more or less your paper to achive a lighter or darker version of your negative. When you make adjustments you are reinterpreting the raw data, to give it a "flavour". Exposition at some extent, is just one of this flavours.

But if the exposure is really clipped, you can not recover detail because beyond that limit you do not have information at all.

Take a look at this post: What's the point of capturing 14 bit images and editing on 8 bit monitors?

Answer 2

Exposure can only be corrected as long as there is som information left in the raw data.

You can make a nearly white pixel somewhat less bright, and stretch the difference between some almost equally bright pixels (increase contrast), but of course, if the data is clipped (pure white), all you can do is "invent" something by interpolating neighbouring pixels. (Some highlight recovery algorithms do this when only some colour channels are clipped, leading to unnatural colours)

That said, RAW files have a wider latitude for these corrections, since they usually contain more data than JPEG images (12,14,or even 16 bits per channel vs. only 8 in JPEG). Every RAW conversion must map these values into the narrower range when producing an 8bit output image, and exposure correction can shift the selected data up or down, while other corrections (contrast, highlight/shadow recovery) influence the stretching or compression on different parts of the range.

Answer 3

Jpg is a compressed format where many pixels are simply discarded based on the jpg quality and compression algorithm used.

It is possible that the original RAW (i.e. ALL pictures taken by digital cameras) capture contained pixels which allow the recovery possible. Discarded pixels are lost. RAW 'development' are simply showing them, not creating them.

One of the suggestions made by experts shooting RAW is expose to the right, as it is easier to recovers highlights versus blacks, where there are NO pixels to recover from.