While the torsion test is an excellent tool, it would be useful to have a test that more directly relates to the sharpness of images coming out of the camera.  I am going to duplicate the razor blade mtf test developed by Jim Kasson.  This test takes an image of a single back-lit razor blade, as shown below:

The razor blade transmits no light, and so there is a very high contrast level between the blade and the white surface behind it.  In addition, the edge of the razor blade is very precise, ensuring that we have the sharpest possible target to be imaging.  Any less perfect subject, such as a printed black line on paper, would result in a less sharp edge for the camera to image and thus an apparent decrease in sharpness.

Once the image is captured, we will calculate the MTF 50 over the edge of the razor blade.  I am not going to rehash a full description of MTF here, but the internet is full of resources on the matter.  In essence though, we are asking how imaged edge is.  The sharper the image, the higher the MTF.  MTF 50 is the spatial frequency of alternating black and white lines that is blurred by any imperfections in the camera/lens/tripod to only have 50% of their original contrast.  If the camera is moving around due to improper support, only large structures in the image have any hope of maintaining the original contrast, and the image will appear blurry.  MTF 50 is simply a quantitative metric for asking how small of structures will still appear sharp.

To calculate the MTF 50, I use the free program MTFmapper.  The output of MTFmapper looks as such:

The program is calculating the MTF over the edge and reporting the result in blue text.  The calculated number is in cycles per pixel.  The spatial resolution being resolved on the sensor at 50% contrast is thus

$pixel width / (cycles/pixel)$

Given that the maximum possible is 0.50 cycles/pixel due to aliasing, a value of 0.36 is quite good.  This particular image was taken with a Pentax 645z and 120mm FA lens.  The bayer color filter will also decrease the maximum attainable resolution.  It is important that the input to MTF mapper be unsharpened for consistent and meaningful results.

Here is another example, taken with a Canon 6D II and 85mm @ F/5.6.

This is also an excellent lens and it shows.

In future posts I will explore exactly what effect a good or bad tripod can have on the sharpness of the image using MTF 50 as the quantitative measure.

## 4 thoughts on “Razor Blade MTF test”

1. John says:

Would it not be a lot simpler to use
a backlit USAF 1951 test slide? I
have used one from Edmund Optics for
resolution tests on birding scopes.
Mine shows horizontal and vertical
patterns ranging from 1 line pair/mm
to 14 line pairs/mm, which should be
suitable distance.

1. David Berryrieser says:

I can’t plug the images from that into MTF mapper. Presently I am just using a printed MTFmapper target as opposed to the razor blade since I am more concerned with differences in resolution than an absolute MTF measurement.

1. Doug Kirk says:

It seems to me that the ink bleed from the printer could be a limiting factor for the absolute sharpness
value. But if you’re only interested in relative measurements, then so be it.

1. David Berryrieser says:

Yes, the resolution of the printer could be a factor. I just try to make sure that I don’t get too much magnification on each square by backing up sufficiently for each lens. But yeah, I am not looking for absolute MTF values. Just trying to figure out where the tripod could degrade sharpness.