MTF Weight Calculations

By: twinsemi 5326 days ago
Concerning your weighting algorithm for various apertures, do you factor in the swept area for the data collected? I was looking comparing the data for two Canon EF lenses, the 50mm f/1,4 and f/1,8. (using the OLD site, I brought up two charts and compared them side-by-side) Looking at the f/8 chart for the f/1,8 lens, I noticed a significant drop in performance between 12mm and 18mm. The f/1.4 lens did not show this, but its weighted MTF at that aperture was only ,01 better than the f/1,8 lens. (0,86 vs. 0,85) I must conclude that the weighting favors the center of the FOV.

From a statistical point of view, the center resolution is almost meaningless. The area of the film or sensor for the region between 0 and 1mm is pi*(r1^2- r2^2) = 3mm^2. (or one times pi)
For the radii from 11 to 12mm, which is still entirely within the 35mm frame, the area is 23 times pi. (72mm^2)
Beyond 12mm, the math is a little too tedious for 99,99999% of the population. All right, I confess. I have forgotten the formulas, and am just too lazy to look them up on Wikipedia.

Maybe a case could be made for a heavy weighting of the center region at wide open, since we do a larger share of our focusing there. But then all of the current AF systems use a variety of focusing points, so that falls apart.

Maybe I am missing something in the graphical interpretation, but I am not a newcomer to statistics, graphing, or photography. From the looks of it, the Canon 50mm f/1,8 lens should not rate anywhere near the f/1,4. (it is currently 4,2 vs. 4,4, respectively) What do you think?

This all started last week, when I got my first camera with an APS format sensor. Now I have to look at the charts, since so many of the lenses have real performance differences at the fringes. How about a second set of numbers for the APS side of the street? (like you don't have enough else to do, right)
By: johnriley 5326 days ago
It's technically interesting but in reality APS-C sized sensors are using the centre of the FOV and getting the best from all lenses. For many film-designed lenses this is not enough and colour abberations spoil things, but most of the good marque lenses provide excellent real-world results.

Digital-design telecentric lenses do give superior results in many cases, but then they are designed for the purpose.

Testing procedures are all well and good, but in the end it's the photographic results that count. It'll be interesting to see the technical answert though!


Add your message

Login required
Please login here or if you've not registered, you can register here. Registering is safe, quick and free.