SMI is a form of contrast enhancement developed by Jerry Lodriguss . The technique is particularly suited to enhancing the visibility of faint detail. It uses a screen combine with the bright areas covered by a mask made from the inverted luminance channel of the image.

Let's start with an image that could benefit from enhancement of some of the dim detail.

The algorithm proceeds as follows:

1. First duplicate the image and call the top layer A and make it the active layer
2. Use a gaussian blur with a one or two pixel radius
3. Make a layer group including layer A
4. Make a mask from the inverted luminance channel of the blurred layer as the upper layer of the group
5. Set the combine mode of the group to screen

The layer stack looks like the figure below.

The mask layer is shown below.

Finally flatten the image stack. If the background is too light a levels adjustment layer can be used to adjust it to taste. The processing can be applied multiple times to increase the effect. The final image is shown below.

Finally after repeating the whole process a second time we get the image below. Compare this to the original by moving your mouse over the image. Note the dim nebula between the Pelican and NGC7000 is now visible and the image doesn't suffer badly from star bloat.