• Members 1433 posts
    May 31, 2025, 12:05 a.m.

    Interlacing...

    I had the word acuity in my mind then forgot it.

    I was referring to pixels / inch or line pairs / mm or whatever is used to define what is necessary to achieve the appearance that someone wants.

    I keep saying to myself there is no point getting involved with these word / interpretation games.

  • Members 299 posts
    May 31, 2025, 6:58 a.m.

    You don't control the resolution in modern cameras, it is set by the system, and as resolution is set in the output media the "appearance" is limited to zoom, especially with modern printers who's algorithms do a far better job of interpolation than anything a user can achieve.

    Ahh, I really shouldn't get involved in these in depth discussions as often ideas are summarily dismissed...

  • Members 1426 posts
    May 31, 2025, 9:15 a.m.

    Thanks to Arvo for providing the answer already :-)
    While such transformations into the frequency domain are mathematical, an appreciation of how they look in the frequency domain, and how different response functions and signals combine, can help one understand the total system response and the final result.
    There are also many practical uses of using FFT, DFT,...

  • Members 872 posts
    May 31, 2025, 3:29 p.m.

    Yes, for looking at texture, I use the GIMP Threshold function set to 0.5 on the magnitude plot out of G'MIC's FFT.

    tractor no adj.jpg

    tractor no adj FFT.jpg

    I use the Threshold function so that all frequencies over 50% "power" come out pure white as can be seen above, especially useful to me for comparisons. This plot tell me that "MTF50" is sort of occurring around 0.2cy/px because the edges of the plot represent Nyquist, and the original image was taken in 2x2 binned mode on-sensor, i.e. 10 um pixels.

    tractor no adj FFT.jpg

    JPG, 1.6 MB, uploaded by xpatUSA on May 31, 2025.

    tractor no adj.jpg

    JPG, 1.8 MB, uploaded by xpatUSA on May 31, 2025.

  • Members 872 posts
    May 31, 2025, 4:10 p.m.

    I wondered if a computer monitor screen might have an MTF similarly to Norman Koren's work on scanners:

    It seems like it might, if one accounts for the magnification factor and views at 100% zoom:

    monitor MTF.jpg

    Human Visual Acuity excluded from the above Open Office spreadsheet.

    Typo in I11 should say H6/30, not H7.

    <snip>

    P.P.P.S. found definitions of printer and monitor MTF - see Section 3.3 - they include human visual acuity!

    corp.dxomark.com/wp-content/uploads/2017/11/2012-CPIQ_Spacial_Metrics_Copyright.pdf

    monitor MTF.jpg

    JPG, 423.7 KB, uploaded by xpatUSA on May 31, 2025.

  • Members 872 posts
    June 1, 2025, 9:52 p.m.

    AI just gave me this:

    typ monitor MTF curve.png

    So if I have a camera with a sensor pixel pitch of 5 um then that is a spatial frequency of 100 lp/mm. A 100 ppi monitor has a spatial frequency of about 100/25/2 = 2 lp/mm per the above chart, in other words a magnification factor of about 50.

    typ monitor MTF curve.png

    PNG, 106.1 KB, uploaded by xpatUSA on June 1, 2025.

  • Members 872 posts
    June 3, 2025, 7:08 p.m.

    Of all the measurements of "resolution" I prefer Acutance as implied by the Edge Spread Function (ESF) at right:

    FS.jpg

    DPR test image, Sigma sd Quattro H, QuickMTF slant-edge.

    FS.jpg

    JPG, 1.0 MB, uploaded by xpatUSA on June 3, 2025.

  • Members 872 posts
    June 5, 2025, 3:33 a.m.

    According to Falk Lumo, a 10 to 90% rise of 1.26 px is optimal - any more tends toward soft and any less implies over-sharpening.

    www.falklumo.com/lumolabs/articles/sharpness/index.html Section 1.2.2

  • Members 2431 posts
    June 5, 2025, 4:14 a.m.

    all i know is shooting extreme macro at 10x pixel size needs to be between 5um and 10um for max resolving power. how does that relate to your charts.

  • Members 872 posts
    June 5, 2025, 4:34 a.m.

    Yes, we all know that you shoot extreme macro, Donald. I have no idea how that relates to the conventional resolution charts posted in this thread.

    Anybody?

  • Members 872 posts
    June 5, 2025, 4:04 p.m.

    I have a vague memory from prior work as an Electrical Engineer. Convert the polar FFT to Cartesian co-ordinates. Multiply the rectangular values. Convert the result back to polar and apply the inverse transformation?