Ted Miller

Vlad, You said " First point. Here I understood everything, my assumption turned out to be correct and I can safely make relative measurements without changing the test conditions. I mean not changing the distance between the test target and the camera. Everything should be permanently in place, I can only change lenses. Yes, I realized that if something more than relative comparison starts to interest me, I have to be very careful about the quality of the test chart." In a general sense what you...

Vlad, You will get "the most accurate measurement data possible" only after you have spent many hours practicing and increasing the stability and repeatability of your test setup. I can speak from experience, as learning to make MTF measurements has turned into my COVID project. I did not intend it to be my project, but it took over my COVID time. I have exposed over 1500 images of MTF targets since March, and I can now begin to trust the results I am getting, because I get repeatable results, and...

I noticed that the heading for column 4 in the '-v 2' version of edge_mtf_values.txt seems to be hard-coded to say MTF50, when in actuality the value reflects the MTFxx value requested on the command line or in the GUI box. Could you please make the header refect the actual MTFxx value. It can be very confusing to try to figure out that what is actually recorded there is an MTF90, which comes out with a very strange value for an MTF50 calculation. (Don't ask me how I know.) Ted in Indiana, USA

In case it wasn't clear, all of what Frans included in his answer is available from the command line (and a lot of it is not in the GUI). Ted in Indiana, USA

After compiling MTFM in a virtual machine, I started the program. It came up in the virtual console and looked OK. Then I clicked on Settings / Preferences. The screen was only set to 1024x768. The main window fit OK, but the Preferences box overflowed by quite a bit, and there is no way to scroll it. Even with the screen height set to 960, it still overflows so much that I cannot see the ACCEPT button. The only way to change preferences was to go to another computer and reverse engineer the tab...

Frans, I decided that I wanted to civilize this script before turning it loose on the public. I put some explanation at the top, and some other things. I also found that the script did not install MAKE. Of course that made CMake throw an error. I had a brand new Centos 8 Virtual Machine that I had set up, so I used the script in it. After some tweeks it runs OK, though I wish I could easily try it on a blank Centos 8 computer, to see if I missed anything. I am attaching the updated version. You have...

My package installer says that opencv 3.4.6 is installed on my Centos 8 laptop. I get no answer to pkg-config for opencv2 and opencv4, but # pkg-config -cflags opencv -I/usr/include/opencv The OpenCVConfig.cmake file is located at: /usr/lib64/cmake/OpenCV/OpenCVConfig.cmake I put the line OpenCV_DIR="/usr/lib64/cmake/OpenCV" before the cmake line, but I still get the error message. I don't know if this is relevant, but I noticed that when I execute locate opencv | grep include only these files show...

It was an unmosaiced image. This one wasn't the one that made me look for the problems, but I will keep in mind that unmosaiced images are likely to have ripples in them.