Canon MTF charts

[mjhoward]

Does anyone know what LP/mm the Canon MTF charts were evaluated at?

 

[mjhoward]

nobody? I guess I might try calling Canon tech and see if someone there knows.

 

[Studio7Four]

Canon's MTF charts have several series of lines. I believe from what I've read on Canon's website in the past (they've moved things and I can't find the link now) that they generally present the MTF with the lens wide open and then at its sharpest.
The sweet spot is different by lens model, and likely even varies somewhat by unit for a given model, which is why the charts are intentionally obtuse (the plots aren't labeled).

I find the MTF charts to be a useful generalization of the performance of a lens, but I wouldn't get too hung up on specifics unless you have a specific application (beyond just taking photos) for which you really care about optical performance.

 

[cfusionpm]

If your goal is to compare lenses (specifically Canon lenses) and the kind of images they can produce, this website has a fantastic tool. It gives visual examples at pretty much every focal length and every aperture for just about every Canon lens.

Canon EF 70-200mm f/4.0 L IS USM Lens ISO 12233 Resolution Chart Results

 

[mjhoward]

Canon's MTF charts have several series of lines. I believe from what I've read on Canon's website in the past (they've moved things and I can't find the link now) that they generally present the MTF with the lens wide open and then at its sharpest.
The sweet spot is different by lens model, and likely even varies somewhat by unit for a given model, which is why the charts are intentionally obtuse (the plots aren't labeled).

I find the MTF charts to be a useful generalization of the performance of a lens, but I wouldn't get too hung up on specifics unless you have a specific application (beyond just taking photos) for which you really care about optical performance.

I believe you are correct in that the difference between the blue lines and the black lines are the difference in one aperture and another. However, I think that the thick lines and thin lines represent the results taken at different resolutions... or line pairs per millimeter (LP/mm) to eliminate sensor size as a factor. The charts will look completely different for sensors with different pixel pitch because optics have a limit as to what size 'spot' they can resolve. The MTF for a lens used on a 5D will look MUCH MUCH better than the same lens on a 7D body because the pixels are much larger and are able to be resolved by the lens better.

And yes, I have a specific application in which I care about optical performance. I only need the Canon specs as a point of reference against a C-mount lens for machine vision.

 

[Studio7Four]

Gotcha. I still haven't been able to find the explanation on Canon's website, but doing a little search got me this: Re: How to read CANON's MTF chart: Canon SLR Lens Talk Forum: Digital Photography Review. It looks like the answer to your question is that the thick lines are at a spatial frequency of 10 lp/mm and the thin lines are at 30 lp/mm.

Sorry if my first response seemed like I was belittling your question. I was assuming (perhaps errantly) that most people on here don't want/need to know about the nitty gritty of optics.

 

[mjhoward]

That would actually be good (for my situation) if 10LP/mm and 30LP/mm are what is indicated on their charts but it wouldnt make a whole lot of sense as far as DSLR's go. The formula for determining LP/mm is 1/(2*pixel size in mm). So for their current line-up, I believe the body with the largest pixel is the 1DMKIII with a pixel pitch of 7.2um. This means the lowest resolution limit of any of their current sensors is about 70 LP/mm. Most of their sensors now have a limit of at least 116 LP/mm. As you can see by their charts, the smaller the pixel (higher tested LP/mm), the less the lens is able to resolve it and the worse the chart looks.

I was just a bit worried because I received an MTF chart on a machine vision lens, and while it had a good transfer function, it was only evaluated at 12 LP/mm. The sensor I would like to use it with has resolving limit of 145 LP/mm.... which is why I was really interested in some form of reference to something I'm a bit more familiar with.