New CiC Tutorial: Camera Lens Quality - MTF, Resolution & Contrast

[McQ]

This is an "advanced topics" tutorial, on par with the existing article about diffraction.

Warning: this tutorial can at times get technical

Camera Lens Quality: MTF, Resolution & Contrast

It gives an overview of concepts relating to lens quality, contrast and resolution, in addition to how to read and compare MTF charts. There's also a little on the "sweet spot" of a lens when it comes to sharpness/contrast. It's quite long compared to the last couple I've added, so there's lots to read...

As usual, any corrections/feedback/suggestions/etc are welcome.

[Colin Southern]

"Everyone is likely familiar" -> "Everyone is likely to be familiar"?

[McQ]

"Everyone is likely familiar" -> "Everyone is likely to be familiar"?

Thanks, this has been changed. The article is now accessible from the main tutorials page...

[KentDub]

Wow... I think my brain hurts...

Maybe add another paragraph or two about reading the MFT charts (Maybe a review, or alternate way of saying the same thing)... I understood it going though the article, but if I poped up Canons site right now I wouldn't be able to read them...

[McQ]

OK, thanks for the feedback on that section. I'll go through and rework the "How to read an MTF chart" section a little

Update: I've added two more paragraphs near the beginning of the "how to read an MTF chart" section. Let me know if this helps, and if there's something you feel is still missing. Thanks.

[KentDub]

I think it's better, thank you. Still trying to memorize all of it I don't want to be a pain or anything, so feel free to ignore this. There are so many combinations of lines (colors, thickness, stroke) -- that maybe a cheat-sheet table/matrix would be helpful in addition to the great descriptions you have. "If you want a lens that is good for this situation.... Look at this line" -- or the opposite.

Something else I was confused on. For a full-frame sensor I understand that you read the full graph (as it goes to 22.5mm - for a crop-sensor how far horizontally do you read the graphs on the horizontal axis? If it is different, maybe a combination of the sensor sizes overlaying the graph would be a nice image to add.

Just wanted to say thanks for adding another tutorial though - I always find them helpful.

[arith]

These tutorials are invaluable. I'm looking for a decent lens and now know it has to be prime, since it is not only cheaper, lighter, smaller but sharper as well. (for what I want it for)

cheers

[Colin Southern]

These tutorials are invaluable. I'm looking for a decent lens and now know it has to be prime, since it is not only cheaper, lighter, smaller but sharper as well. (for what I want it for)

cheers

Keep in mind though that "sharper according to the MTF charts" and "visibly sharper" are two very different things.

Correct sharpening technique will have a much bigger influence on the outcome than any inherant difference between any two reasonable quality lenses.

[McQ]

These tutorials are invaluable. I'm looking for a decent lens and now know it has to be prime, since it is not only cheaper, lighter, smaller but sharper as well. (for what I want it for)

Prime versus zoom lenses is another topic altogether

As Colin mentioned, MTF is definitely not the end all be all of lens considerations. While a prime might be a little sharper (although in the MTF tutorial the zoom does admirably), a zoom lens is much, much more versatile. When you consider that you might have to crop the photo more often with a prime (because you cannot get into just the right position for that focal length), any image quality differences even out. You'll need to move around a lot more, and change lenses a lot more (potentially adding more dust to the sensor), in order to use a prime effectively. Both of these can mean that you'll take more time to get the shot. For me personally, I find that a prime can be very restricting when light is changing fast and I want to try a lot of different compositions. I end up changing lenses far too often, when I would have otherwise been taking other photos...

[Colin Southern]

Prime versus zoom lenses is another topic altogether

@ Arith

Hi Arith - out of interest, we had a good discussion regarding lenses and a bit of "prime" -v- "zoom" not too long ago in this thread here.

I think Sean said it very concisely (as usual). Personally in the Prime -v- Zoom debate I tend to think of;

- Size being of no consequence (one may be smaller, but only by a handful of millimeters)

- Same for weight - only a few ounces in it

- Sharpness - the same for all intents and purposes other than 100% tests.

- Usability - this is where the zoom wins BIG TIME, and in my opinion offsets all of the other factors by a big margin.

Which is interesting since I also own several primes! (that I also don't use a lot I might also add). The biggest joy of primes for me is their speed (in that I can explore the range between F2.8 all the way down to F1.2 with the likes of the EF85mm/1.2L.

[arith]

Cheers, but too late now since I paid for a used ef 28mm f2.8 apparently in mint condition. It is small and therefore I think better suited to panorama, has good reviews and mostly I'm stuck on 28mm occasionally 35mm but not much.

I still have the other lenses for snapshots; I would have liked a zoom and found that the 28-105mm MKI although Canon don't mark them as such, would have been ideal.

But none for sale and the posty is on strike soon.

Cheers.

[Alis]

Hi Sean,

I was reading this one (very interesting topic by the way) and noticed that under MTF, 5th paragraph, in this sentence:

A high-end lens with an MTF-50 of 50 LP/mm will appear far sharper than a lower quality lens with an MTF of 20 LP/mm, for example (presuming that these are used on the same camera and at the same aperture; more on this later).

you actually mean:

...with an MTF-50 of 20 LP/mm...

Not sure though. Just thought mention it.

Cheers,

Alis

[McQ]

Yep, thanks Alis - that second MTF should read MTF-50. It's been changed.

[David]

McQ - This is another excellent tutorial with the complex matter being explained clearly and succinctly compared to other versions on other sites. I'm sure there are going to be more calls for all these tutorials and more to be put into hard copy!

Cheers

David

[Dave Humphries]

Hi Sean,

Eventually I have found the time to give this a good read
and it is a GOOD read as usual.

Also "as usual", I hereby present my overly pedantic/thorough set of thoughts

All of the different looking lines in the above MTF charts can at first be overwhelming

(only at first?)

When trying to assimilate the charts and the keys (Bold, dashed, thick, thin, etc.), I found I was doing a lot of scrolling up and down the screen, I can't think of a better suggestion at the moment...except maybe...

I do wonder whether having eight bits of info on one graph is just too much (?).
Thinking of the psyscology (sp?) of presentation and how a large percentage of readers can hold say, 5 things in their short term memory (e.g. in terms of bulleted list lengths), but the more items you add the less % of people can handle it. (Am I teaching granny about eggs?)

Is it worth separating the Bold (pop) and Thin (Fine res) aspects to two separate pairs of graphs?
I suggest the split there because it is helpful to see the two apertures overlaid and also the Meridional and Sagittal differences.

I'm still trying to think of a memorable analogy for those two Can't we just stick to "radial" and "concentric"? The other names might be more accurate (and deserve a mention), but they were completely new words to me and they don't aid my understanding, whereas I already know what radial and concentric mean.

I liked the green concentric circled graphics demonstrating the radial dimensions and how these delineate what's irrelevant on the MTF-50 graphs above.

However, overall, my initial, and even now, perception of the MTF-50 graphs is that the zoom (on left yes?) looks better than the prime (on right?). I draw this conclusion because the left results are generally closer grouped and higher (esp. for the central area) than the right hand graph. However, that's not the textual conclusion you have drawn, so I must have missed something
Or it's the 'overload factor' affecting my ageing brain

Moving on...
I am slightly puzzled by the inclusion (in the first of the closing bulleted lists) of "image noise" - the other four points are very obviously relevant to lens quality alone and if considered would help you differentiate between lenses. Image noise is surely only relevant to the camera/sensor test rig isn't it?

Other quality reducing facors might also consider:
- IS/VR used inappropriately? (e.g. on tripod)
- in fact, doesn't Image noise belong in this list?

"Micro abrasions, moisture, fingerprints or other coatings on the front of your lens", and not if on the rear element? I suggest removing the italicised bits

As Colin has said above, is it also worth saying that a good PP sharpening technique can have dramatic effects outweighing the MTF-50 differences? Perhaps add as a third element to the closing paragraph?

[Dave Humphries]

One other thought regarding those MTF graphs and keys; how about if the graphs were one above t'other and have the key beside?

[will_c]

Personally in the Prime -v- Zoom debate I tend to think of:
- Size being of no consequence (one may be smaller, but only by a handful of millimeters)
- Same for weight - only a few ounces in it
- Sharpness - the same for all intents and purposes other than 100% tests.

As a counterexample,

50mm 1.4 : weight 290g, length 50.5mm
24-70 2.8L : weight 950g, length 123.5mm

At 50mm, at equivalent apertures, the 50mm is at least as good. Plus it can go two stops faster. The 24-70L is more than four times as expensive. So as long as one mainly uses around about the middle to longer part of the range one might well go for the 50mm. If the wide angle side is less important, and you're not going to photostitch, you could buy one of the better wide angle compacts and still come out way ahead as far as expenditure goes.

To me the main reason for going for primes, unless you have bucket-loads of money to spend, is to get quality and speed at a reasonable price (I don't have shares in 'Primes Inc'). But this has all been said before.

[McQ]

To me the main reason for going for primes, unless you have bucket-loads of money to spend, is to get quality and speed at a reasonable price (I don't have shares in 'Primes Inc'). But this has all been said before.

A great point that deserves emphasis. I completely agree. You get an amazing amount more for your money (quality-wise) by going with a prime...

[McQ]

Thanks Dave for all the analysis. See below:

I do wonder whether having eight bits of info on one graph is just too much (?).
Thinking of the psyscology (sp?) of presentation and how a large percentage of readers can hold say, 5 things in their short term memory (e.g. in terms of bulleted list lengths), but the more items you add the less % of people can handle it. (Am I teaching granny about eggs?)

Is it worth separating the Bold (pop) and Thin (Fine res) aspects to two separate pairs of graphs?
I suggest the split there because it is helpful to see the two apertures overlaid and also the Meridional and Sagittal differences.

I agree with you here 100%. An MTF chart would be much clearer if it were broken up into separate charts. However, this is just one of those things where that's how MTF charts are shown. While I don't agree with having such a complicated chart, in a section on "how to read an MTF chart" I ought to show them what they will be actually be seeing.

On another note, I suspect that the complicated look to the chart is intentional. It makes the lens look very scientific and overwhelms the reader, so that they eventually say "oh, ok, it must be good."

I'm still trying to think of a memorable analogy for those two Can't we just stick to "radial" and "concentric"? The other names might be more accurate (and deserve a mention), but they were completely new words to me and they don't aid my understanding, whereas I already know what radial and concentric mean.

I think I will just mention sagittal & meridional once; after that I will use radial and concentric throughout. The only disadvantage to this though is perhaps the case where someone googles sagittal or meridional lines, and wants to see an explanation. The site will be much less likely to provide them with that if the words are each mentioned once out of ~7 pages of text. Searches are not the end all be all goal though...

However, overall, my initial, and even now, perception of the MTF-50 graphs is that the zoom (on left yes?) looks better than the prime (on right?). I draw this conclusion because the left results are generally closer grouped and higher (esp. for the central area) than the right hand graph. However, that's not the textual conclusion you have drawn, so I must have missed something

Yes, all the lines can be quite a mess The key is that the black lines with the lens wide open. That's why the prime lens *looks* worse than the zoom at first glance, when in reality we're comparing apples with oranges. No matter how good the lens, it will be very hard pressed at f/1.4 (as with the prime) to beat out another lens at f/2.8 (as with the zoom).

If you only look at the blue lines, which correspond to both lenses at f/8.0, then the prime clearly wins here.

I am slightly puzzled by the inclusion (in the first of the closing bulleted lists) of "image noise" - the other four points are very obviously relevant to lens quality alone and if considered would help you differentiate between lenses. Image noise is surely only relevant to the camera/sensor test rig isn't it?

Yep, it's been removed. I was originally going to include lots of other quality-reducing factors unrelated to optics, but never expanded on it. To keep things simple, I'm going to just remove that and have the first set of bullets refer only to optical properties, instead of also including IS/VR, etc.

"Micro abrasions, moisture, fingerprints or other coatings on the front of your lens", and not if on the rear element? I suggest removing the italicised bits

Done.

As Colin has said above, is it also worth saying that a good PP sharpening technique can have dramatic effects outweighing the MTF-50 differences? Perhaps add as a third element to the closing paragraph?

I'll add something at the end about post-processing.

Thanks again...

[Dave Humphries]

I agree with you here 100%. An MTF chart would be much clearer if it were broken up into separate charts. However, this is just one of those things where that's how MTF charts are shown. While I don't agree with having such a complicated chart, in a section on "how to read an MTF chart" I ought to show them what they will be actually be seeing.
~
I think I will just mention sagittal & meridional once; after that I will use radial and concentric throughout. The only disadvantage to this though is perhaps the case where someone googles sagittal or meridional lines, and wants to see an explanation. The site will be much less likely to provide them with that if the words are each mentioned once out of ~7 pages of text. Searches are not the end all be all goal though...

Hi Sean,

Glad to help.
Your comments have prompted two ideas that may address these problems (I hope).

On the first para; if you have the time and inclination, I envisage one of your excellent mouse-over graphics, where the default shows a whole MTF chart, but you can move the mouse around to hide certain aspects by hovering over captions like;
Show f8 only - removes all wide open (black) lines
Show wide open only - removes all f8 (blue) lines
Show concentric (Meridional) only - removes all radial (solid) lines
Show radial (Sagittal) only - removes all concentric (dashed) lines
Show all
Thus each option halves the amount of info being assimilated and flicking between captions often aids understanding.
I think leaving the 10 and 30 Lp/mm always on together is probably wise, although an alternative would be to separate these and show concentric and radial together, but I think we'd learn more as I first suggested.

One more thought; if f8 is more important and comparable than wide open, given the different apertures, why is f8 a lighter colour than wide open, I'd suggest reversing, so the more obvious lines are the more important ones - or is this another of those daft (IMHO) convention things?

On the second para; How about having a CiC Glossary of photographic terms?
Or just use this form; "concentric (Meridional)" everywhere?

As I say, I'm glad to help, thanks for taking these onboard,