Using Moire to see if the lens out-resolves the sensor

This post is aimed mainly at those who sometimes use test charts to see how good their lenses are. All that really matters is whether the lens resolves as much as the sensor. Using a lens that resolves a lot more than the sensor will give very little improvement over using a lens that is just good enough.

I have occasionally tested my lenses with test charts and, by accident, I noticed that certain types of test chart often produce coloured moire fringes even though the chart is purely black and white. This effect is very obvious; it is not necessary to pixel-peep the images to see how well each different line spacing is resolved.

The moire fringes are produced only if the image contains regular detail with a spacing very approximately the same as the pixels in the sensor. If the lens isn't capable of resolving that fine detail in the image, then moire fringes will be absent or very faint.

So the presence of well-defined moire fringes indicates that the lens is as good or better than the sensor.

The test pattern that I have found works best is:

Images of this with a good lens show Moire fringes (the weaker the AA filter on the camera, the stronger the fringes should be):

GM1 with Sigma 60/2.8 at f/4 (this is a very small crop at full magnification)

GM1 with Sigma 60/2.8 at f/8

At f/8, diffraction blurring is significantly weakening the Moire fringes. If the image is slightly out-of-focus, the fringes disappear and the disc becomes a uniform grey, as they do if the lens fails to produce a sharp image.

For this test to work, the diameter of the disc should be around 160 pixels in the image, but the beauty of this test image is that it does not need to be sized accurately. It will still work at 200 pixels wide (or more) and at considerably less than 160 pixels.