How to steepen the contrast
When a yet unprocessed 16-bit TIFF image is loaded for the first time, it often appears black with a few white dots. In order to make the buried parts of the image visible, one has to iteratively steepen the contrast curves (Image -> Adjustment -> Curves):
Depending on the image depth and the dynamic range selected, I do between 3 and 8 such iterations.
When to stop steepening the contrast (noise limit)
I usually stop when the background level reaches a value of about 20 in R,G,B. This is when also the noise becomes visible, but depends on the dynamic range chosen.
The background level can be measured with the eyedropper tool, setting the sample size to 5x5 average to avoid confusion by noise. The values can be found in the Info window:
Avoiding saturation
In particular galaxies have a very bright core and faint outer haloes. To show both, one must exclude the brighter inner regions from further contrast steepening. This can be done by means of a feathered Lasso tool. The region of an image which would saturate (or become too bright) during the next steepening of the contrast curves can be selected with the lasso tool. In order to avoid discrete jumps or hard edges in the image, the boundary of the region selected should be diffused. This can be achieved by feathering the lasso. The feather radius is given in pixels and should approximately have the same size as (or be smaller than) the diameter of the region selected.
Several regions can be selected at one time with different feather radii. Finally, the selection gets inverted (Select -> Inverse) and then the curves are steepened:
The selected region remains unaffected, while the fainter outer parts are enhanced. The transition region is not visible.
Other methods to avoid saturation
The same effect can be achieved using a nonlinear scaling of the entire image, such as DDP, logarithmic or else. However, the method with the lasso tool described above offers most flexibility and highest control and is preferred by me. Yet, it is cumbersome.