Jump to content

Slanted Edge Technique


Recommended Posts

Advertisement (gone after registration)

Testing of lenses and images is quite simple with the Slanted Edge Technique.

 

Take a picture of a high contrast slanted black/white edge and cut out a part at 100%, like this

 

 

(don't worry about the white balance, the method corrects for it)

 

Now run this image through sfrmat3, a software package written for Matlab (but there are also independent packages available) and you get this:

 

 

This is the Modulation Transfer Function (MTF) or Signal Frequency Response (SFR) of that part of the image.

 

You can see three color channels (r, g and

and the Luminosity (L), so that is the black and white component of the image.

 

It was made with a Nokton 50/1.1 lens at f/1.1 on a Leica M9 with a sensor resolution of 3699 ppi.

 

You can see that the modulation is lower for the blue channel, because of the combined spherical and chromatic aberrations.

 

Furthermore you can see that the half-sampling resolution (Nyquist frequency) is still just represented by the lens, so the sensor is limiting at 75 linepairs/mm

 

So now it is quite easy to compare lenses by comparing graphs and you can see if this agrees with the data from the manufacturer.

 

Along the vertical axis is the modulation (1 is perfect, 0 is no information) and along the horizontal axis the detail in the image in terms of spatial frequency in line pairs/mm. The half_sampling point is the maximum the sensor can produce. The higher the curve, the better the lens.

 

Do this without sharpening of the RAW file to get a proper reading of the modulation (sharpening enhances modulation, but in a non-propertional way along the frequency axis).

 

With this method you can also see if your focussing was adequate by comparing different settings: The one with the highest curve, stretching out further to the right is the best. You just need a simple image of a slanted edge in black and white with high contrast. It is important that this edge is close to perfect: straight and super sharp.

 

Also you can establish the optimal aperture of a lens with this method, for instance for architecture or document photography.

 

Here an explanation of the method: Imatest - MTF curves and Image appearance

 

You can find the MatLab version here: sfrmat2, sfrmat3, SFRedge: slanted-edge analysis for digital cameras and scanners

 

A simple and superfluous warning on this Leica site: This is just a limited signature of the properties of a lens. It just says how contrasty a lens is and what the resolution limit is. All the other important properties are not represented. So use this method as just a partial finger print of the lens :-)

Edited by Lindolfi
added a warning
Link to post
Share on other sites
Interesting indeed. No Mac version?

 

Matlab is available for Mac as shown here: MathWorks Nordic - MATLAB - Requirements

 

However, Matlab is very expensive for personal use. There is a student edition, and you could try and find a friend who uses Matlab (student, researcher, R&D person) to get some help at accessing and running Matlab on a temporary basis.

 

I use Matlab professionally, but at home I use Octave which is one of several free Matlab clones. There may be minor compatibility problems depending on the code in the sfrmat3 script, but usually they can be fixed.

 

Other possibilities are (not exhaustive):

 

Sage: Open Source Mathematics Software

 

PyLab -

 

Home - Scilab WebSite

 

Regards

Per

Link to post
Share on other sites
...

 

With this method you can also see if your focussing was adequate by comparing different settings: The one with the highest curve, stretching out further to the right is the best. You just need a simple image of a slanted edge in black and white with high contrast. It is important that this edge is close to perfect: straight and super sharp.

...

 

Did you compare this directly to the usual kind of focus test?

 

Cf. http://www.l-camera-forum.com/leica-forum/customer-forum/209621-tolerances-summicron-28mm-2.html#post1888349 where someone else used a contrast method.

Link to post
Share on other sites

The test with a target chart at an angle with the sensor plane is quicker, but the slanted edge method is precise and if you need it for another purpose (like getting the SFR curve), it can be combined in one experiment.

 

If I set out to check focussing I use the angled chart method you refer to, #12.

 

By the way: it is surprising how good the human eye can see differences in contrast, so just "eye-balling" charts is not such a bad idea

Link to post
Share on other sites

Thanks for the post, that's interesting.

 

I was wondering if a through-focus test with this is more precise than just shooting a ruler. Did you make your own target, or buy one? Anything like this for Mathematica, not Matlab?

Link to post
Share on other sites

Advertisement (gone after registration)

Yes, I made the target by cutting a piece of black film and glue it on white glass.

The source code for MatLab is open. If you have experience writing for Mathematica, perhaps you can translate the code after asking the authors permission. The code is not that long and perhaps it outweighs buying MatLab.

Edited by Lindolfi
Typo
Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...