Strange values on vertical edges
Status: Beta
Brought to you by:
fvdbergh
Menu
▾
▴
Hi, Frans
A photographer friend of mine who is fond of testing photographic equipment is very interested in this program, but he had one question. He asked me to ask the author of the program. The question is:
"On edges close to vertical there are some unrealistic values. I don't know what is connected with. Probably, vertical ones should not be analyzed at all ... He knows better. Imatest analyzes edges inclined by 5%"
Regards,
Vlad
Last edit: Vlad_spb 2020-10-11
Hi Vlad,
Correct. Edges that are perfectly vertical or horizontal should not be used for slanted-edge analysis. You will notice that the MTF Mapper test charts do not contain such edges when oriented properly.
But MTF Mapper will allow you to process images of arbitrary test charts, so it is up to the user to be aware of this when using non-MTF Mapper test charts. As you can see in the examples, those vertical edges are annotated in red- or yellow-, rather than cyan-coloured text to indicate that these measurements are not ideal. Edges annotated in red are ignored in most of the higher-level MTF Mapper outputs (e.g., the "grid" output).
There is a new measurement in recent GUI versions of MTF Mapper (say, 0.7.30) that directly shows you just how bad a particular edge angle is. If you click on the annotation over the edge (the red/yellow/blue text) in the GUI's display of the "Annotated" output, the SFR dialog will show you, amongst other things, a column titled "OSF" in a table. If you take your "MTF mapper_4.jpg" example above, you can compare the OSF values of the slanted (~ 5 degrees) edges to the vertical edges. The slanted edges should give you an OSF of 8.0, whereas the vertical edges will be much lower, maybe 2.0 or even lower. (hover over the table column in the SFR dialog to see some tooltips on interpreting the values).
By the way, "OSF" is an acronym for Over Sampling Factor. The reason why we use a slanted edge in these measurements is to effectively sample the edge at a higher sample density than what you would expect to be possible using the camera sensor. I wrote a blog post on that here .
Lastly, your third example above "MTF mapper_5.jpg" is not really suitable for the automated analysis mode of MTF Mapper. Let us say you tilted the chart a little so that the vertical edges of the "i" were at a 5-degree angle. The probem is still that the automated analysis mode of MTF Mapper (the default "File->Open" option) will not correctly process the edges of the "i" because of the serif marks (I think that is what they are called) that cause the "i" to not be a simple rectangular shape. There is actually a beta version of MTF Mapper (0.7.31) that you can download https://sourceforge.net/projects/mtfmapper/files/windows/mtfmapper-0.7.31-win64.exe/download that has a preview of a new manual edge selection option (File->Open with manual edge selection). If you are careful to select only the simple region of the edge (i.e., exclude the serif marks) then you can actually use the edges of the "i", after tiltining the chart a bit, of course. But note that this new feature is still under development. I hope to complete this feature within the next month or so.
Regards,
Frans
Thank you very much Frans!
I will give your answer to a friend. He wanted to use your program to write articles on testing new lenses. Your information is very important to him. Of course, I'll take a closer look myself.
Regards,
Vlad