Canon EF 24-70mm f/4L IS MTF Chart comparison

Home · Register · Search · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda

Canon EF 24-70 f/4L IS for $1,299

Pre-order new Sigma 18-35mm f/1.8 DC HSM lens

Bower 35mm f/1.4 price dropped to $379

FM Forums \| Canon-mount SLRs \| Join Upload & Sell
1 2 3 4 end

Canon EF 24-70mm f/4L IS MTF Chart comparison

AGeoJO
Offline
• • • • • •
Upload & Sell: On

p.3 #1 · Canon EF 24-70mm f/4L IS MTF Chart comparison

S Dilworth,
Thanks for the breakdown!

Nov 13, 2012 at 12:35 AM

S Dilworth
Offline
• •
Upload & Sell: Off

p.3 #2 · Canon EF 24-70mm f/4L IS MTF Chart comparison

mmurph wrote:
I have to admit though that I didn't bother to sort out all of the lines and charts this time!

I don’t blame you. Each of those tiny MTF charts has eight curves, mostly overlapping. Maybe Canon pays by the square inch for its website?

Keep in mind I only teased out the data for the extreme corners (image height 21.6 mm) at the 24 mm focal length. The lenses perform more similarly (though far from identically) across much of the frame.

At longer focal lengths everything might change again. I haven’t really checked (and there’s no 70 mm curve for the 24-105mm).

Nov 13, 2012 at 12:36 AM

jcolwell
Offline
• • • • • •
Upload & Sell: On

p.3 #3 · Canon EF 24-70mm f/4L IS MTF Chart comparison

S Dilworth wrote:
I don’t blame you. Each of those tiny MTF charts has eight curves, mostly overlapping. Maybe Canon pays by the square inch for its website?

Ziess MTF diagrams are about the same - with respect to how much screen space they occupy....

Nov 13, 2012 at 12:42 AM

Yohan Pamudji
Offline
• • •
Upload & Sell: Off

p.3 #4 · Canon EF 24-70mm f/4L IS MTF Chart comparison

S Dilworth wrote:
mmurph wrote:
I have to admit though that I didn't bother to sort out all of the lines and charts this time!

I don’t blame you. Each of those tiny MTF charts has eight curves, mostly overlapping. Maybe Canon pays by the square inch for its website?

Conserving bandwidth to save money. Next: coffee rationing in the office.

Nov 13, 2012 at 01:52 AM

RobertLynn
Offline
• • • • • •
Upload & Sell: Off

p.3 #5 · Canon EF 24-70mm f/4L IS MTF Chart comparison

skibum5 wrote:
RobertLynn wrote:
Great, I own out dated 1 series, and both the 24-70, and 24-105.
If my clients find out, I'm finished!

No seriously though, this shit always happens. Lol.

I've actually had my ear to the gossip mill lately and I did hear that many of your clients were just shocked, SHOCKED, when they realized you were using the 24-70 I and 24-105 and will never even bother so much as returning another call or email ever again and will look the other way should that happen to pass you out and about. Most have already burned and deleted (and by deleted I mean removed the contaminated HDs without even bothering to rescue anything else on them and tossed them into the test fusion reactor in Princeton). I think I even saw a WANTED poster up with your mug on it.

I knew it. Those pricks.

Nov 13, 2012 at 02:24 AM

Fred Miranda
Offline
Admin
Upload & Sell: On

p.3 #6 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Going back to the MTF charts for a moment, I disagree with earlier suggestions that the 24-70mm f/4L IS is comparable in performance to the 24-70mm f/2.8L II. The f/2.8 lens has much stronger corner performance according to these charts.

Concentrating on the sagittal lines alone, the 24-70 II @24mm has the contrast/resolving power edge wide-open and at f/8. However, at 70mm, things reverse and the 24-70 f/4 IS does a little better.

I’ve pointed out where the 24-105mm actually does better than the new f/4 lens, but overall the new lens is clearly better.

The meridional line advantage you pointed out is obviously irrelevant but it's nice because it may confuse us into thinking our 24-105L @24mm wins in something!

Like you wrote, we still need to compare other aspects not revealed by the theoretical charts like vignetting, distortion, CA, flare susceptibility and color.

Nov 13, 2012 at 02:26 AM

mco_970
Offline
• • • •
Upload & Sell: On

p.3 #7 · Canon EF 24-70mm f/4L IS MTF Chart comparison

S Dilworth wrote:
Going back to the MTF charts for a moment...

Thanks for sorting out the information in the MTFs.

My wallet is bummed to hear the results, though.

Nov 13, 2012 at 04:00 AM

AJSJones
Offline
• • •
Upload & Sell: On

p.3 #8 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Fred Miranda wrote:

Like you wrote, we still need to compare other aspects not revealed by the theoretical charts like vignetting, distortion, CA, flare susceptibility and color.

If I recall - the MTFs are calculated (nothing wrong with that as some basis for comparison across Canon's lime) but using only green light - we'll have await Samuel's analysis of those curves as a function of wavelength

For the technically inclined here is an excellent pdf from Zeiss on "How to read MTF curves" - if you survive that here is part 2 of the saga.

Nov 13, 2012 at 04:14 AM

Steve Spencer
Offline
• • • • •
Upload & Sell: On

p.3 #9 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Looking just at the extreme corners for MTF graphs probably doesn't make a lot of sense. Still the numbers reported in this post aren't the right ones:

Here they are reported correctly:

24 mm, full aperture, 10 lines/mm, sagittal
24-105mm f/4L IS: 27 %
24-70mm f/4L IS: 89 %
24-70mm f/2.8L II: 97 %

24 mm, full aperture, 10 lines/mm, tangential
24-105mm f/4L IS: 18 %
24-70mm f/4L IS: 24 %
24-70mm f/2.8L II: 48 %

24 mm, full aperture, 30 lines/mm, sagittal
24-105mm f/4L IS: 15 %
24-70mm f/4L IS: 25 %
24-70mm f/2.8L II: 51 %

24 mm, full aperture, 30 lines/mm, tangential
24-105mm f/4L IS: 10 %
24-70mm f/4L IS: 15 %
24-70mm f/2.8L II: 51 %

And stopped down to f/8:

24 mm, f/8, 10 lines/mm, sagittal
24-105mm f/4L IS: 72 %
24-70mm f/4L IS: 98 %
24-70mm f/2.8L II: 100 %

24 mm, f/8, 10 lines/mm, tangential
24-105mm f/4L IS: 50 %
24-70mm f/4L IS: 70 %
24-70mm f/2.8L II: 87 %

24 mm, f/8, 30 lines/mm, sagittal
24-105mm f/4L IS: 66 %
24-70mm f/4L IS: 62 %
24-70mm f/2.8L II: 92 %

24 mm, f/8, 30 lines/mm, tangential
24-105mm f/4L IS: 59 %
24-70mm f/4L IS: 55 %
24-70mm f/2.8L II: 81 %

At 70 the new f/4 lens is even better.

S Dilworth wrote:
Going back to the MTF charts for a moment, I disagree with earlier suggestions that the 24-70mm f/4L IS is comparable in performance to the 24-70mm f/2.8L II. The f/2.8 lens has much stronger corner performance according to these charts.

Canon’s MTF charts ignore diffraction, and they might assume other unrealistic conditions (Canon doesn’t tell us), so they can’t be trusted to give a full picture of lens performance. But they might still provide some clues as to how various Canon lenses compare. Unfortunately the charts have far too much information in a tiny space.

To make some of the info more digestible, below I’ve taken the MTF value for the extreme corner of the frame, for three lenses, at full aperture and f/8, for 10 and 30 lines/mm, for sagittal and tangential orientations, at the 24 mm focal length.

So these numbers compare corner performance at the wide-angle setting, which is a common preoccupation.

First, here are the numbers at full aperture, which is one stop faster for the f/2.8 lens. The performance gap between the f/2.8 lens and the f/4 lenses would increase if the f/2.8 lens were stopped down to f/4!

24 mm, full aperture, 10 lines/mm, sagittal
24-105mm f/4L IS: 50 %
24-70mm f/4L IS: 71 %
24-70mm f/2.8L II: 88 %

24 mm, full aperture, 10 lines/mm, tangential
24-105mm f/4L IS: 58 %
24-70mm f/4L IS: 54 % (worse than the 24-105mm)
24-70mm f/2.8L II: 81 %

24 mm, full aperture, 30 lines/mm, sagittal
24-105mm f/4L IS: 17 %
24-70mm f/4L IS: 23 %
24-70mm f/2.8L II: 48 %

24 mm, full aperture, 30 lines/mm, tangential
24-105mm f/4L IS: 10 %
24-70mm f/4L IS: 15 %
24-70mm f/2.8L II: 52 %

And stopped down to f/8:

24 mm, f/8, 10 lines/mm, sagittal
24-105mm f/4L IS: 72 %
24-70mm f/4L IS: 98 %
24-70mm f/2.8L II: 100 %

24 mm, f/8, 10 lines/mm, tangential
24-105mm f/4L IS: 65 %
24-70mm f/4L IS: 61 % (worse than the 24-105mm)
24-70mm f/2.8L II: 82 %

24 mm, f/8, 30 lines/mm, sagittal
24-105mm f/4L IS: 20 %
24-70mm f/4L IS: 88 %
24-70mm f/2.8L II: 98 %

24 mm, f/8, 30 lines/mm, tangential
24-105mm f/4L IS: 14 %
24-70mm f/4L IS: 27 %
24-70mm f/2.8L II: 52 %

I’ve pointed out where the 24-105mm actually does better than the new f/4 lens, but overall the new lens is clearly better. However, the new f/2.8 lens is much better again. Much better (for sharpness in the extreme corners according to these MTF charts).

Nov 13, 2012 at 05:21 AM

S Dilworth
Offline
• •
Upload & Sell: Off

p.3 #10 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Steve, looking at the corners alone is insufficient for a complete picture, but people are widely concerned with corner performance at the wide-angle setting, presumably because it’s often poor in this type of zoom lens.

I hear fewer complaints about the long end, since it's usually better, and the corners are often out of focus at longer focal lengths (in portraits, etc.).

In any case, the Canon MTF charts have too much information to absorb at once, so we have to concentrate on one aspect or another of performance to take in anything at all.

Back to the corner numbers, then. I’d expect two people to read the charts slightly differently, perhaps by a few per cent in some cases, but some of your numbers are very different from mine. Were you using the MTF charts published on the Canon USA website here, here, and here?

Alternatively, are you sure you’re interpreting the eight lines on the charts correctly? Here’s the decoder from EF Lens Work (PDF):

This image is copyrighted by the owner

And here's a batch for which we got different numbers:

Steve Spencer wrote:
24 mm, f/8, 30 lines/mm, tangential
24-105mm f/4L IS: 59 %
24-70mm f/4L IS: 55 %
24-70mm f/2.8L II: 81 %

According to EF Lens Work, the curve in question for the above data (tangential (or meridional) MTF at f/8 and 30 lines/mm) is the dashed, blue, thin line. For that line, for the charts on Canon USA, I get the values I stated earlier:

24 mm, f/8, 30 lines/mm, tangential
24-105mm f/4L IS: 14 %
24-70mm f/4L IS: 27 %
24-70mm f/2.8L II: 52 %

But if I’ve gone bonkers from staring at these tiny charts, please put me straight!

Nov 13, 2012 at 09:16 AM

S Dilworth
Offline
• •
Upload & Sell: Off

p.3 #11 · Canon EF 24-70mm f/4L IS MTF Chart comparison

I should add that I’m not someone who demands top corner sharpness from every lens. On the contrary, I’m often attracted to smaller and cheaper lenses that sacrifice a bit of corner sharpness, or fast lenses that have unusually good centre performance at full aperture, but never really sharpen up in the corners.

Nonetheless, for technical work like architecture, landscape, etc., reliable corner sharpness at the wide end can be useful. It looks like Canon’s new f/2.8 zoom does very well in this regard. I’m only surprised that Canon managed to do that while significantly reducing the size and weight compared to the previous 24-70 mm f/2.8 lens.

I guess $2300 goes a long way towards achieving the very difficult!

Nov 13, 2012 at 09:25 AM

Ralph Conway
Offline
• • • •
Upload & Sell: Off

p.3 #12 · Canon EF 24-70mm f/4L IS MTF Chart comparison

S Dilworth wrote:
Steve, looking at the corners alone is insufficient for a complete picture, but people are widely concerned with corner performance at the wide-angle setting, presumably because it’s often poor in this type of zoom lens.

I hear fewer complaints about the long end, since it's usually better, and the corners are often out of focus at longer focal lengths (in portraits, etc.).

In any case, the Canon MTF charts have too much information to absorb at once, so we have to concentrate on one aspect or another of performance to take in anything at all.

Back to the corner numbers, then. I’d expect two people to read the charts slightly differently, perhaps by a few per cent in some cases, but some of your numbers are very different from mine. Were you using the MTF charts published on the Canon USA website here, here, and here?

Alternatively, are you sure you’re interpreting the eight lines on the charts correctly? Here’s the decoder from EF Lens Work (PDF):

This image is copyrighted by the owner

Nov 13, 2012 at 09:33 AM

alundeb
Offline
• • • •
Upload & Sell: Off

p.3 #13 · Canon EF 24-70mm f/4L IS MTF Chart comparison

S Dilworth wrote:
Keep in mind I only teased out the data for the extreme corners (image height 21.6 mm) at the 24 mm focal length. The lenses perform more similarly (though far from identically) across much of the frame.

A little more detail on that: The new zooms seem to perform similarly out to about 17 mm image height (almost out to the left and right edges), after that the F4 has a steeper curve out to the corners. If someone wants to crop the image to a square aspect ratio, the F4 lens is perfect for that

Nov 13, 2012 at 11:45 AM

phuang3
Offline
• • •
Upload & Sell: Off

p.3 #14 · Canon EF 24-70mm f/4L IS MTF Chart comparison

It seems the 24-70/f4 version has more image curvature than 24-70/f2.8 II. The later is indeed a superb lens at wide open! It's probably the best std zoom lens on the market.

Nov 13, 2012 at 12:40 PM

Steve Spencer
Offline
• • • • •
Upload & Sell: On

p.3 #15 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Hi Scott,

I see now that you just used the key that you referenced and I just reasoned which lines were which based on first principles and used the graphs on the first page of this thread. I am used to looking at actual measured MTF versus theoretical MTF, however, and that threw off some of my judgments. Basically if you measured the MTFs for f/2.8 or f/4 zooms then sagittal/tangential pairs should be in this order. The top pair should be f/8 10mm, then wide open 10mm, then f/8 30mm, then wide open 30mm. If the key you provided is correct, which I assume it is, then the theoretical graphs suggest that the f/8 30mm line is higher than the wide open 10mm line, which would be implausible with measured MTF.

There is another important aspect to the MTF graphs as well. Normally you would expect sagittal/tangential pairs to be close to each other and follow a fairly similar curve across the graph although these two sets of lines would typically diverge toward the right of the graph (i.e., in what would be the corners). Note this is the way the lines look in your key. This is an important aspect of MTFs because the extent that these lines diverge, the lens is likely to show lateral chromatic aberrations. So based on first principles, I was assuming that blue and black in the graph was indicating sagittal vs. tangential rather than f/8 vs wide open. Now that you have provided the correct key, what is particularly striking is how separated the sagittal and tangential lines are. None of the lenses are great, but the two new lenses are notably better and should have less lateral CA.

One last point is about thinking about the corners. It is probably not best to think of the far right of the graph as the corners. The extreme right is actually only the last pixel or so in the corners. Corner performance is probably best understood as from about 18mm to 21.6mm from the centre on the x-axis. With this in mind we see that the new f/2.8 is clearly the best at 24mm, but the new f/4 lens falls about half way between the new f/2.8 and the old f/4 lens.

I am sorry, I didn't look at the key and therefore misread some of the numbers, but I hope this discussion is helpful. In general, it would be much better if Canon would provide real measured MTF graphs as these theoretical graphs differ significantly from what they would actually be in practice.

S Dilworth wrote:
Steve, looking at the corners alone is insufficient for a complete picture, but people are widely concerned with corner performance at the wide-angle setting, presumably because it’s often poor in this type of zoom lens.

I hear fewer complaints about the long end, since it's usually better, and the corners are often out of focus at longer focal lengths (in portraits, etc.).

In any case, the Canon MTF charts have too much information to absorb at once, so we have to concentrate on one aspect or another of performance to take in anything at all.

Back to the corner numbers, then. I’d expect two people to read the charts slightly differently, perhaps by a few per cent in some cases, but some of your numbers are very different from mine. Were you using the MTF charts published on the Canon USA website here, here, and here?

Alternatively, are you sure you’re interpreting the eight lines on the charts correctly? Here’s the decoder from EF Lens Work (PDF):

This image is copyrighted by the owner

Nov 13, 2012 at 12:48 PM

S Dilworth
Offline
• •
Upload & Sell: Off

p.3 #16 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Steve (I’m Samuel, by the way, not Scott), I’m not sure that theoretical versus measured MTF would particularly matter. In Canon’s case, the curves ignore diffraction (hence their frequently impossible values), and they might incorporate other odd assumptions: it’s hard to say, because Canon doesn’t tell us anything about the focus criteria or spectral distribution of the modelled light.

It would be trivial (for Canon, not me!) to model the full performance in software and come up with an MTF chart that accurately describes real-world performance. I don’t know why Canon (and Nikon, etc.) doesn’t do that, other than perhaps to achieve MTF-chart consistency with old lenses for which the MTF curves were calculated by modelling geometric optics alone.

I don’t find it universally implausible for the f/8 curves for 30 lines/mm to be higher than the full-aperture curves for 10 lines/mm (especially sagittal). You just need a lens that improves strongly upon stopping down. If your point is that f/2.8 and f/4 zooms of the type we’re talking about don’t typically improve that much, I’d agree.

It’s risky to estimate lateral chromatic aberration from the difference between sagittal and tangential MTF. A better indication might be had from the improvement in off-axis tangential MTF when the lens is stopped down. But in this case, since we know neither the type of light Canon uses for the MTF calculations (this is critically important for chromatic aberration), nor the focus criteria, I wouldn’t read much into it.

Steve Spencer wrote:
It is probably not best to think of the far right of the graph as the corners. The extreme right is actually only the last pixel or so in the corners. Corner performance is probably best understood as from about 18mm to 21.6mm from the centre on the x-axis. With this in mind we see that the new f/2.8 is clearly the best at 24mm, but the new f/4 lens falls about half way between the new f/2.8 and the old f/4 lens.

Agreed.

Steve Spencer wrote:
In general, it would be much better if Canon would provide real measured MTF graphs as these theoretical graphs differ significantly from what they would actually be in practice.

I’d be content if Canon merely told us the light and focus criteria used, and modelled diffraction.

Nov 13, 2012 at 03:13 PM

Steve Spencer
Offline
• • • • •
Upload & Sell: On

p.3 #17 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Hi Sam, (sorry I got your name wrong)

I don't think we disagree at all. If you have a f/1.4 lens then yes the stopped down 30mm lines might well be higher than the wide open 10mm lines, but with f/2.8 or f/4 zooms this isn't very likely. Similarly I agree that sagittal/tangential differences aren't the best indicators of lateral CA, but when the discrepancy grows as you stop down CA is at least a plausible source of the pattern. Anyway, thanks for helping clarify charts.

Nov 13, 2012 at 04:30 PM

S Dilworth
Offline
• •
Upload & Sell: Off

p.3 #18 · Canon EF 24-70mm f/4L IS MTF Chart comparison

And thanks for your input. Are you planning to get the lens?

Nov 13, 2012 at 04:43 PM

Steve Spencer
Offline
• • • • •
Upload & Sell: On

p.3 #19 · Canon EF 24-70mm f/4L IS MTF Chart comparison

I'm note sure if I will get the lens or not. I like the lens, but not the price and for my style of shooting I generally prefer primes to zooms, so I'm not sure I would use it enough to justify the price. My wife, however, prefers to shoot with zooms, so we'll see. At a minimum I would wait at least a year to see where the price settles.

Nov 13, 2012 at 05:05 PM

jorkata
Offline
• • •
Upload & Sell: Off

p.3 #20 · Canon EF 24-70mm f/4L IS MTF Chart comparison

Steve Spencer wrote:
Corner performance is probably best understood as from about 18mm to 21.6mm from the centre on the x-axis.

A FF sensor is 36mm wide, so anything beyond 18mm on the MTF chart is actually outside of the frame (and hence irrelevant).
I'd consider 15-18mm on the MTF charts as extreme corner, whereas 12-15mm can be considered 'edge'.

Nov 13, 2012 at 05:27 PM

1 2 3 4 end
FM Forums \| Canon-mount SLRs \| Join Upload & Sell 1 2 3 4 end