Question about the new DXO site and lens measurement methods

[Johnny]

I noticed that dxomark.com has new features, you will now find lens measurements data for lenses and camera/lens combinations. I find the measurements data very confusing. Maybe someone can explain it to me?

Question: What is the difference between lp/mm and lw/ph ?

If I look at the Canon 85mm f/1.8 on a 50D body photozone.de has measured it to ca 2500 lw/ph at f:5,6 which is portrayed as close to the sensor limit. If I look at the same lens/body at the new dxomark site it is measured to 42 lp/mm which is portrayed as average. A closer look at the "Resolution" tab makes it more confusing, showing colors from red to green and only full frame bodies are able to hit good (green) levels even if stopped down. It is probably my lack of knowledge but I find the site and the way the results are portrayed very confusing.

I tried to search Bob´s site but could not find any lp/mm equivalent from his testing of the 85mm.

I went to the dxomark site because I wanted to check out certain lens/body combinations. I now think I will forget about that until I better understand the data.

Any comments are appreciated!

- Johnny

[Bob Atkins]

It's confusing and I'll take a closer look at it.

lw/ph (line widths per picture height) is more applicable as a measure of digital sensor resolution rather than lens resolution. It's origins lie in measuring the resolution of analog TV screens.

lp/mm (line pairs per mm) is more appropriate as a measure of lens resolution - but not when measured using a relatively low resolution digital sensor. For meaningful numbers you have to remove sensor resolution from the equation which you can do using a technique using a slanted edge target rather thana traditional line pattern. It involves measuring the sharpness of an edge then doing a fourier transform on the data. Sounds complex (and it is) but in practice it's pretty easy with the right software. I can do it but I generally don't since it's more of academic then practical interest.

To futher add to confusion, 1 line pair = 2 line widths!

With a 15mm high sensor, 2500 lp/ph corresponds to 166 lime widths/mm. This would then correspond to 83 lp/mm.

I don't know how DxO are defining their "resolution" numbers or what their testing methodology is, but I'll take a look at the new site and see if I can figure out what they are doing. It might make a good subject for an article here.

Note that they're not testing lenses, they're testing [Lens + Camera] combinations. They also appear to be scaling resolution numbers based on sensor size so the resolution numbers they quote would be proportional to the resolution in a print of a given size, not necessarily to a screen image or the image as captured on the sensor or the resolution of the lens itself.

More later...

[Johnny]

Thank you, Bob. I will check back later.