{"id":31044,"date":"2022-01-27T22:26:14","date_gmt":"2022-01-27T21:26:14","guid":{"rendered":"https:\/\/phillipreeve.net\/blog\/?p=31044"},"modified":"2023-06-20T13:06:33","modified_gmt":"2023-06-20T12:06:33","slug":"vignetting-graphs-explained","status":"publish","type":"post","link":"https:\/\/phillipreeve.net\/blog\/vignetting-graphs-explained\/","title":{"rendered":"Vignetting Graphs Explained"},"content":{"rendered":"

Introduction<\/h2>\n
\"\"
Sigma 35mm 1.2 Art DG DN<\/figcaption><\/figure>\n

In our reviews we usually give figures for the vignetting of the lens and thanks to a reader with programming skills I got the opportunity to extend this and show you some fancy graphs how the vignetting changes when stopping the lens down. You will get the fancy graphs, but there are some caveats we need to talk about first.<\/p>\n

<\/p>\n

Measuring Light Falloff<\/h2>\n

So what we ususally do in our reviews is this: we take shots of the overcast sky with a white tissue over the lens and then we adjust the exposure in Lightroom for the corners to match the brightness (in terms of the RGB values) of the center of the frame.
\nGenerally with this method we saw a good match with the vignetting figures e.g. Lenstip derived with their Imatest software or opticallimits is publishing, so we didn’t question this method. Now when assembling these graphs we made some discoveries that overcomplicated things significantly.<\/p>\n

\"\"
Relationship between RGB values and Light intensity (EV)<\/figcaption><\/figure>\n

The first issue is that the relationship between the RGB values and the vignetting – measured in intensity or “EV” – is non linear. We tried to overcome this with a so called fit-function that maps the EV values to the RGB values. For that we took pictures of an evenly lit surface with varying exposure (EV) and compared those changes in exposure to the corresponding RGB values.<\/p>\n

\"\"<\/p>\n

Then we discovered another issue, depending on the overall brightness in the picture the tonal curve changes. So this fit-function was only usable for a certain (and too small) dynamic range. The following graph shows the evaluation of 9 pictures which are all 1\/3 EV apart.<\/p>\n

\"\"
Graphs should overlap, but the brighter pictures deviate a lot<\/figcaption><\/figure>\n

The follow up issue was that even when trying to stay within this range the EV values in these graphs gave out values very different from those we derived with our previous method or could find elsewhere.<\/p>\n

I will use the Sony FE 85mm 1.4 GM<\/a> as an example. At f\/1.4 I measure 2.2 EV vignetting in the extreme corner (I use the Raw file with Adobe Standard processing). Lenstip measured a mean value across all 4 corners of 2.26 EV<\/a> (Lenstip is using OOC jpegs), these values are very much comparable.<\/p>\n

\"\"
Sony FE 85mm 1.4 GM<\/figcaption><\/figure>\n

When using the aforementiond fit-function it gives out a value of 1.7 EV (see graph above) which is too big of a departure from the other values for me to publish it.<\/p>\n

The “Solution”<\/h2>\n
\"\"
Sony FE 85mm 1.4 GM<\/figcaption><\/figure>\n

So what I will show you instead is a graph which I will call “brightness”. This is a rather unscientific name that Lenstip is also using for a similar graph in some of their reviews, e.g. that of the Sony FE 85mm 1.4 GM<\/a>.<\/p>\n

This brightness graph is merely based on the RGB values and not to be mistaken<\/strong> with the light intensity graphs you find e.g. in Zeiss or Leica datasheets.<\/p>\n

\"\"
Sony FE 85mm 1.4 GM<\/figcaption><\/figure>\n

How does it work? The script is simply getting the red values along the diagonal and normalizing them, so an RGB value of 0 becomes a brightness value of 0 and an RGB value of 255 becomes a brightness value of 1. All the lines for different apertures are adjusted, so that the maximum value (to be found in the center of the frame) is always equal to 1.<\/p>\n

At this point it shall also be noted: Zeiss and Leica are giving vignetting values for the lens alone, when we measure such things (or try to) we always measure the system which consists of lens, sensor, in camera processing and post processing (like raw development). We try to keep these effects as small as possible, still what we measure will always be higher than the values that manufacturers give for the lens alone.<\/p>\n

What our graphs cannot do:<\/h2>\n