Measuring picture quality objectively?!

[Crodol]

Is there a way to measure the quality of an image objectively. The general case might be quite complex but maybe there is a practical way for the classic case that we encounter - a (hopefully) sharp bird on a washed out background.

The only things I can think of is the size of the object in pixels (the more the better :-) and the "number of unique colours" which is sometimes useful when I cannot decide which of two very similar pics is better.

Any thoughts - or links to prior discussions - would be appreciated.

[Mono]

I suppose there are many objective ways of measuring a digital image, and artists have mused for millennia on the qualities of a perfect picture. But at the end of the day it is the subjective view that matters be it yours, a magazine picture editor's or a competition judging panels.

[RAH]

One method used to judge the sharpness of images is to view them at 100% - i.e. zoom in with the image editor/viewer to a 100% size, which is also called "pixel peeping." When you do this while viewing 2 images together, it is fairly easy to judge which is sharper. This is why folks post "100% crops" on websites sometimes to show sharpness. A 100% crop means that no downsizing was done to the image - you have just cropped out a small section from the original image and are displaying it.

[njlarsen]

I forgot the details, but there is something about the size of an automatically generated jpg. If two images are from the same distance, same composition, but one is sharp and the other blurry, then the jpg will differ is size for the two situations (I think the sharp one is larger, but you have to check for yourself).

Niels

[RAH]

I think that I've noticed that jpgs files get a little larger after enhancement - e.g. sharpening, contrast enhancement, etc. But in a way you'd think the opposite would be true, because blur should use more colors, not less. For example, it's a bad thing to use gradients (one color gradually turning into another) or even anti-aliased lines and characters (non-jagged edges) in web images if you want them to be as small as can be, because these techniques add additional colors.