Canon 16-35 f/4L IS USM MTFs vs the 16-35 f/2.8L

In Reviews On Wednesday, May 14, 2014

This article compares the theoretical performance of the Canon 16-35 f/4L IS USM with the 16-35 f/2.8L II by looking at the MTF graphs for the two lenses.

Full Review of the 16-35 f/4L

Introduction

The Canon EF 16-35 f4L IS USM was announced yesterday and it will have a lot of landscape photographers considering upgrading their current wide-angle zoom lens. There were previously just 2 zoom lenses in the 16-40mm range, the 17-40 F4L and the 16-35 F2.8L II. I have owned both and used them extensively. Optically there is little if anything to choose between them in terms of sharpness and contrast when stopped down. The 16-35 produces beautiful 14 points sun stars and has a 2.8 aperture, helpful for astrophotography. The lenses also have a slightly different focal length, but for 90% of landscape shots the images they produce are almost indistinguishable. The 2.8L is similarly sharp at f2.8 as the 17-40 is at f4. Both lenses ‘suffer’ from average edge resolution and soft corners even when stopped down, though in my opinion this is rarely an issue. Nevertheless, that remains the main flaw with these lenses and since the release of some better Nikon alternatives Canon shooters have been clamoring for an upgrade.

MTF curves are theoretical lens performance graphs which, if the real life performance matches the graphs, tell us quite a lot about image sharpness and contrast. Since I currently use the 16-35 f2.8L as my primary (only) wide-angle zoom I decided to make some comparisons with the MTF curves of the respective lenses. As mentioned above the 17-40 is very similar so the results below can be read across. Remember this is ‘just for fun’. MTF curves are theoretical, I assume they do actually stand for something, but I think you have to read the information below with a pinch of salt….

I’m only comparing lenses at 16mm because it takes a long time to do these comparisons and write this blog. I’m also more interested in the 16mm performance because its the focal length I use most but trust me that the MTF at 35mm is equally impressive, perhaps more so.

 

Explaining MTF

The below conclusions are somewhat dumbed down, that is to say I don’t explain everything in detail. If you want a better understanding of how I have come to these conclusions have a look at this graphic explaining MTF.

Also understand that I am comparing an f2.8 lens to an f4 lens and so the ‘wide open’ results really shouldn’t be compared directly. It would have been fairer to have the results from the f2.8 lens stopped down to f4, but sadly these curves do not exist.

 

Explaining MTF – its not as complicated as it looks!

EF 16-35 f/4L IS vs. 16-35 f/2.8L II – Contrast

The new f4L appears to have outstanding contrast characteristics across the frame and at all apertures at both 16mm and 35mm. It should be slightly better than the f2.8L. Any value above 0.8 can be considered as excellent and the new lens achieves this throughout with only the extreme corner dropping to ‘very good’ when shooting wide open at f4. As long as the flare characteristics are good (and we have every reason to believe they will be due to developments in lens coatings in Canon’s most recent lenses) then we have a lens with outstanding contrast. The result should be vibrant colours and deep blacks as good as the best Canon lenses. That said, since I have never ever been disappointed the contrast of the f2.8L perhaps this won’t matter too much!

Contrast at f/8 shows the 16-35 f/4L (red) is clearly better than the 16-35 f/2.8L (blue)

 

Contrast wide open shows the 16-35 f/4L (red) is clearly better than the 16-35 f/2.8L (blue) to the point that at the edges of the image it may be noticeable

EF 16-35 f/4L IS vs. 16-35 f/2.8L II – Sharpness

The sharpness of the 16-35 f/4L IS (red) is significantly better than the 16-35 f/2.8L (blue) particularly in the corners according to the MTF

The ‘sharpness’ MTF curves for the EF 16-35 f4L IS are very impressive indeed. In fact they appear to blow the f2.8L out of the water. Perhaps the most exciting results are those stopped down to f8, the kind of small aperture that most landscape photographers use for the majority of shots. At f8 the 16-35 f4L IS shows a noticeable increase in sharpness in the center but a very significant increase in the corners. The sagittal lines (solid) are going to result in the most objectionable blur and it is here that we see the biggest improvement. In fact sharpness in the sagittal direction should be excellent at f8 all the way to the extreme corner. There *should* be a night and day difference, exactly what everyone was hoping for!

The wide open results are of less interest to landscape photographers and no doubt people might read this an say “you can’t compare one lens at f2.8 to another at f4, that’s not a fair fight”. Well that’s true, but here are the results anyway and it looks like the f4L is, as expected, significantly better.

The 16-35 f/4L IS (red) is significantly sharper than the 16-35 f/2.8L (blue) from center to corner, but perhaps that is to be expected given the aperture difference?

 

In fact just for fun here is a comparison between the 16-35 f4L IS at f4 and the 16-35 F2.8L II at f8. The f4L is still better!

The 16-35 f/4L IS (red) is sharper at f/4 than the 16-35 f/2.8 (blue) is at f/8 !

That result is surprising and also seems unlikely! There is only one place I have found reliable samples and that’s the Canon Japan site here: http://cweb.canon.jp/ef/info/ef16-35/index.html Regardless of what aperture these images are taken at (and reading the page it seems the left image may be taken at f/4) these samples look extremely promising!

Samples from Canon Japan. Shame I can’t read Japanese!

EF 16-35 f/4L IS vs. 16-35 f/2.8L II – Astrophotography

The newer f4L lens is a full stop slower at its widest aperture making it less suited to astrophotography. However the MTF curves show that it is appreciably sharper, particularly in the corners. If you compare it to the 17-40L at f4 (which is a bit of a pig for astrophotography) you see exactly the same magnitude of sharpness improvement. Furthermore, comparing the sagittal lines to the meridional lines indicates that there won’t be the same degree of astigmatism in the corners. The f2.8L wasn’t by any means great for astrophotography, but it was good enough. I doubt the f4L will be any better, but the sharpness improvements should make the 1 stop loss in light gathering ability far more acceptable.

 

EF 16-35 f/4L IS vs.……

I thought I would do a couple more f8 sharpness comparisons against the best prime and zoom lens offerings Canon have at the wide end

An MTF comparison of the 16-35 f/4L IS at 16mm and f8 with Canon’s best offerings in this range

If you compare the red (sagittal) solid lines to all the other solid lines you see it is higher up the graph than all of them from left to right. The same is also true of the dashed lines (meridional) except for at the extreme corner where the 14mm and 17mm TSE are slightly better. Essentially the MTF suggests that the new lens is sharper than every other lens on the graph when stopped down to f8. That’s pretty impressive.

 

EF 16-35 f4L IS vs. 16-35 F2.8L II – Unanswered questions

MTF doesn’t tell the whole story. Missing lens characteristics include distortion, chromatic aberration, flare and vignetting. However, given the calibre of recent Canon lenses it is hard to see any of these factors being objectionable for landscape photographers. The greatest unanswered question for me is: What will the sun stars look like? For those who like bombastic shots into the sun (like me) then the 14 point pinwheel sun stars of the 16-35 f2.8L were a huge bonus, I can only hope that then newer lens is as good. One thing we do know is that since it has 9 aperture blades whatever sun star it does produce will have 18 points!

Image stabilisation is a huge bonus for what I do. When hiking it can be a pain to be setup and take down my tripod to take images on route during the day. I actually have no issue whatsoever in pushing my 6D to ISO 400 to get reasonable shutter speeds. Having IS of anything approaching 3 stops will allow me to shoot handheld in almost any situation during the daylight hours as long as exposure bracketing or long exposures are not required. That is a great thing in my opinion!

 

Conclusion

If the MTF curves can be trusted then the 16-35 f4L IS should provide the sharpness upgrade Canon users have been waiting for in their wide-angle zooms and appears to pave the way for a higher resolution sensor. In fact, when stopped down the new lens appears to perform better than some of Canon’s finest prime lenses. Still, these comparisons are theoretical and real life use might show something different. I think I’m going to have to buy one and find out for myself!

 

13 Comments

  1. Hi Alex,

    Interesting stuff. I have been looking to replace my trusty 17-40 for quite a while now. The new 16-35/4 IS will cost the exact same as the 16-35/2.8 II (now with cashback). Which one would you prefer, assuming I am not an astro photographer or some kind of modern day Henri Carter-Bresson. My gut says the 16-35/4 IS, also because it has the same filter thread as my 70-200. Any insights if you were in my position?

    Thanks, and please show us some new work!

    • Hi Marijn,
      Certainly I think it is worth waiting to see some real world samples from the f4IS but if you don’t shoot astro I very much doubt that there is any advantage in getting the 2.8 II. If the MTFs of the f4 IS are correct then it will be much better for landscape than the f2.8 II.
      Alex

      • tobyGreg, I want to say Not quite one of the benefits of the STM bediess the silence is smaller adjustments to achieve focus When a USM lens loses focus it cycles all the way through the range of focus before coming back to focus, STM lens is able to make adjustments in either direction at anytime . Let me test this is the next day or so I don’t have the 24-105 on me but I have several other high-end USM lenses that should work as a good example. Will report back.

  2. Thanks Alex. Not in a hurry, so will wait for real-world examples and decide then what to do.

  3. Thanks for the MTF check outs, yes I too had been looking at them and coming to similar conclusions -and Like you I wait with bated breath for more real world corner examples (I use the 17-40 so they’re mush)

    What I especially will keep an eye out for is the optimum fstop for centre AND corner sharpness (resolution), I am prone to shoot with close elements so as focus stacking is a PITA due to focus shift I hope f11 will be noticably better than the 17-40

  4. Great new addition for Full Frame bodies. Love the MTF charts. Now hoping build and
    CA plus lens flare isn’t bad. construction looks similar to 100L, 24-70 f4. Like the internal focusing and as mentioned alreadly love the 77mm filter threads. Canon could have a winner here, I also forgot to mention how I need a 16 not just 17 on the super wide end. Lovely, I already ordered mine.

  5. Thanks for the test!
    In your chapter “Explaining MTF” unfortunatly there is a major misunderstanding:
    While the y-achsis is correctly referred to the contrast (with 1 is the best and usually normed to 1 on x-scale zero), the explanation of the x-achsis is confusing:
    It´s not the distance from the center but the frequency in linepairs per mm.
    A MTF therefore shows the contrast (or call it “sharpness”) of the image in respect to it´s resolution, the finer structures the more to the right. A MTF chart is only showing the behaviour in one spot of the lens.

    In order to compare the projection quality from center of lens to the border you have to compare multiple MTF charts (if available).

    Regards,
    Martin

    • Hi Martin,

      Thanks for your comment. Unfortunately in this particular case you are mistaken. Although I didn’t want to go into specifics too much on my blog (to avoid scaring people off) different spatial frequencies are actually represented by the thick and thin lines. Canon (and I believe other lens manufacturers) use thick lines to represent measurements at 10 line pairs per mm (‘contrast’) and thin lines represent 30 line pairs per mm (sharpness). The x axis does indeed signify the distance from the centre of the frame.
      However you are correct in asserting that in order to have a full picture of a lenses sharpness reponse at different frequencies you would need a large number of curves. Nevertheless for practical purposes with modern sensors 10 and 30LP/mm gives an excellent representation of real world performance.
      Thanks for the well intended comment.
      Alex

  6. Hi Alex, great article. You can find a sunstar photo here http://www.naturfotocamp.de/?a=blog&id=3886

  7. Hmm looking at photozone looks like the best f-stop for sharpness might be a bit wide open for landscapes (5.6-8).

    I have often thought about focus sacking to get around tis problem but focus shift make sit rathera complicated option

    Mind you Im not that technically minded – there might be a solution I don’t know about to get the absolute best from a lens (yeah I know – get a MF camera – but the weight and costs are more than I can bear!)

    • Yes at f11, there is a miniscule amount of diffraction and it does affect the resolving power of the lens (an almost inperceptible amount). I wanted to test at f11 because that is the aperture I shoot at and also because the corners of the f2.8 II continue to improve to f13.
      On any Canon body focus stacking for sharpness would be overkill. Sharpening would make the difference between f8 and f11 invisible. On a D800 it could be worthwhile if you were really picky!

  8. tobyHello I think I answered your salmiir question elsewhere in quality, chromatic aberration and color balance these two lenses are quite salmiir. The 18-135 with the STM technology does allow for faster and silent AF while filming, it is also silent AF for photography too but the speed of the AF is about the same as USM. Personally I think they range of 18-135 is a little more convenient that the wider 15-85. Hope that helps.

Post a Comment

Your email is never shared. Required fields are marked *

*
*