Resolution

[Kayaker72]

In diving into film, trying to understand the resolution of film, I've started to better understand the different resolution terms and units that I have seen used by different review sites and what they mean.

A few good references, but here discusses the different units used to describe resolution, here is a good write up, and Imatest themselves have one of the best write ups I've seen on how to test resolution, cycles/frequency, MTF, etc and how it applies to photography.

So, just diving into this briefly, common terms are:

• Line pairs (LP): A pair of lines (), but most easily thought of as a black and white line to show contrast.
• Line width (LW): The width of the dark or white line. 2 LW = 1 LP
• Pixels: Pixels are taken fairly idealistically and 1 row of pixels = 1 Line, so the width of a pixel = 1 LW = 1/2 LP
• Cycles: The distance (or time but for photography typically distance) to go from the start of a line pair to the end, so a cycle includes a complete dark line and a complete white line. Thus 1 cycle = 1 LP = 2 LW = 2 Pixel widths
• Dots does get a bit more complicated as dots can overlay each other, but when used for maximum resolution 1 dot = 1 pixel is typical.

• mm: millimeter, so unit of length
• in: inch, starting off easy here, another unit of length. 25.4 mm = 1 in
• PH: Picture height. Everyone treats this as obvious but I've yet to see someone report the picture height even when using the unit.

So, playing with a few values:

• 300 dpi = 300 ppi / 25.4 = 11.8 pixels per mm. Dividing that mm up: 1 mm /11.8 pixels = 0.0847 mm, or 84.7 micrometers/microns. As 2 pixels = 1 LP, 300 ppi is equivalent to ~6 LP/mm. This may sound easily obtainable, but if you are actually only resolving 60 LP/mm on a 36x24 mm sensor, you can only enlarge that image up to 360x240 mm (14x9.4 inches) before your true resolution drops below 300 ppi. 30 lp/mm, and 5x7 is the largest print you can do before dropping below 300 ppi.
• So, I am seeing good film is ~80 LP/mm. Playing with some math, the R5: 8,092 pixel/36mm / 2 pixels/LP = 112 LP/mm. Same math on R6 = 76 LP/mm theoretically possible.
• But, combining the lens on a camera body for an imaging system:
• Lenstips does their resolution reviews and report LP/mm. The highest rated lens using the 5DIII as the base camera is the Sigma 135 f/1.8 at 51.6 LP/mm. Perhaps that doesn't seem like much, as I've seen >80 lp/mm go by. But the 5DIII had 5760 pixels/36 mm, or 160 pixels per mm, and with 2 pixels per lp = 80 LP/mm was the max resolution. So, the sharpest lens lenstip tested achieved about 64.5% of the theoretical resolution of the 5DIII.
• Getting a bit more complicated, but Optical Limits/Photozone. I've always just taken their resolution values as relative and perhaps I should leave it at that. But, they report their values as LW/PH. Which I now know is Line Width per picture height and, ideally, LW = 1 pixel. But, I am not sure about picture height. I could be wrong, but I think it is sensor height. If it is, then a lot of things fall into line.
  
  
  • For example, the latest review of the RF 16 f/2.8 showed peak performance as an impressive 5020 LW/PH at f/4 on the R5. LW to pixels is 5020 pixels/PH, and the R5 has 5464 pixels along its vertical axis. If I can then 5,020/5,464 =~92%. As the sensor height is 24 mm, this also means that the resolution would be 105 LP/mm. If true, impressive in the center....but on the edges, 1968 LW/PH, divide by 5464 = 36% of what was possible, and 41 lp/mm of actual resolution.
  • Optical limits did look at the Sigma 135 f/1.8 on a 5DsR. Peak performance as 5,303 LW/PH, which would be 5,303/5,792 = 92% of what was possible and 110 LP/mm.

I would be very interested if anyone could confirm what Optical Limits means by PH. I did check several different reviews and it does seem like the peak observed at Optical Limits is typically 80-92% of theoretical on each of the different cameras they used (APS-C or FF). So, maybe PH being the sensor height in landscape orientation is the correct assumption, if not, it does seem to correlate somehow.

So, I know this is a bit of a dive, but I am finally putting together some of the math behind resolution, maybe connecting the dots between reviews I've read for years and what it might mean so I thought I'd share.

If correct, you can start to see just how much impact lenses are having on the resolution of camera bodies, very similar to what DXOMark does with their sharpness P-Mpix score.

Whew...ok. I know that was a lot. I don't view this as critical, but it is interesting and potentially helpful in understanding what lens/sensor combinations are capable. Also, the philosophies of digital vs film, digital is trying to make better and better lenses and subdivide the same 36x24 mm area smaller and smaller. So to get high resolution enlargements they need to continue to increase the lp/mm. Film, essentially took the opposite position, lens/film combinations seem somewhat fixed in the 30-80 LP/mm range, but if you want bigger enlargements, increase the size of your film.

[Busted Knuckles]

So that begs the question of sticking a current high res lens on an eos film camera???

And of course diffraction???

[Kayaker72]

So that begs the question of sticking a current high res lens on an eos film camera???

that is in the works.

[Kayaker72]

Having just posted about the MTF of the recent super telephotos, as I've read about resolution, MTF my understanding has changed.

I used to think the y-axis was about transmittance of light. But now I understand it is about shades of gray. 1x at the top would be the blacks and whites as they pass through the lens maintaining the values of the contrast chart (1000:1, etc). So black is pure black and white is pure white a MTF would read 1.

But, as light passes through the lens, resolution is lost and the blacks and whites become muddled together. So, "0" on an MTF is perfectly gray, where you cannot differentiate between the white lines and black lines. All resolution is lost.



This was borrowed from imatest.

So, the black and white lines become blurred as the lp/mm increases, the amplitude, the measured difference of black and white gets smaller (amplitude decreases) you approach gray, the inability of the system to distinguish between black and white lines.

Subtle difference from what I previously thought, but it does make more sense.

BTW, here is Canon's explanation of MTF.

Getting to an actual chart:

The black lines here how much "muddling" has occurred when looking at a chart that has 10 lp/mm. The blue lines are 30 lp/mm. Nice flat lines, that is good, similar resolution across the frame. But, in terms of fine resolution, we are talking about 80 lp/mm for film and the R5 being capable of resolving details at 114 lp/mm. Having lines at 30 lp/mm already 40% muddled (halfway to gray) is not a good sign for resolving power of fine details (IMO).

This all made me recall the "OLAF" testing Roger Cicala/Lens rentals did that is stored on this site. They looked at responses up to 50 lp/mm, here is the Sigma 135 Art against an Otus lens.

Anyway, I know I diving into details here, but writing these things down helps me consolidate my thoughts. So, thanks.

[Joel Eade]

The link to Canon"s explanation of MTF charts leads to a wonderful article that reveals the difference between contrast and resolution.

It goes on to say how these are measured and plotted on the MTF charts as well as a nice explanation of what you can and cannot get from analyzing charts.

Again, thanks for taking the time to post this info.