11.24
I have been surprised at how much interest the first part of this series created, so obviously there were many photographers who wanted to see a direct comparison between the 24-105L and various prime lenses within that range. I am looking forward to see for myself and others how the zoom compares to the 35mm F2.0.
The situation here is identical to the first test with a 1DsII being used on a Gitzo tripod and a cable release to photograph a garden fence with our lenses of choice at F4, F8 and F16. I suggest you read the first post if you want any more details about the methodology for this test and while the subject may lack the sleek lines found in many of my normal subjects, I feel this is a reasonable test purely to see how the lenses stack up for sharpness at different apertures.
As I said in the previous blog post, I have never been hugely impressed by the 28mm F2.8 compared to Canon’s other lenses, but I am expecting the other primes to do somewhat better, purely from what I see in my day to day picture taking and the 35mm is a lens I have used and liked for several years. The reality is that very slight differences one way or the other seen on screen at 100% really shouldn’t bother any you too greatly because those differences will more than likely be lost in printing.
Let’s not waste any more time on chatter though and get straight to the test…
35mm at F4
Canon 35mm F2.0 @ F4 edge sharpness
24-105L zoom on 35mm @F4 edge sharpness
Perhaps not a huge difference here, but I would have to say that the zoom looks better in this situation for sharpness. Both lenses show a little fall off in sharpness at the extreme edges and it would be interesting to see how the 35mm L would look if a comparison were made, but I don’t have that lens and given the amount of wide angle photography I do it’s unlikely that I would invest in one. In practical terms both of these lenses are producing very good results on a full frame camera even at F4. In theory they should both be better at the centre though, so let’s see what happens.
35mm prime @F4 centre sharpness
24-105L zoom on 35mm @F4 centre sharpness
Clearly both lenses have picked up some sharpness at F4, however, it seems fairly evident that the zoom is the sharper lens at F4 which would appear to lessen the argument that faster lenses will automatically be sharper at fairly wide apertures. Certainly at F8 the prime should start to look considerably better, but I am only finding out myself as I write and then view the files side by side at 100%.
35mm at F8
35mm prime @F8 edge sharpness
24-105L on 35mm @ F8 edge sharpness
This result is pretty close and in print probably impossible to see but I feel the prime lens is just a little better here. Common theory would dictate that the prime should be quite a bit better but clearly that L quality counts for something more than just a letter.
On a side note, I can’t help but be struck by the tendency of the L series lens towards a warmer tone and higher contrast in every image seen so far. Whether or not you like that will largely be a case of personal preference, but on a practical issue I always liked the fact in the past that I could switch between my Canon primes and the colour would stay consistent, which is very important on most professional jobs. Obviously corrections can be made during raw processing, but it’s more work. No doubt Canon feel that the look of the L lenses is superior but it does make me think that it’s perhaps not best to mix L series and normal lenses on a job if colour consistency is important.
35mm prime @ F8 centre sharpness
24-105L on 35mm @ F8 centre sharpness
The prime has clearly picked up here a bit compared to it’s performance at F4 as you would expect, but there is definitely a greater rendering of detail from the zoom, so at around the theoretical best aperture for both lenses the zoom proves to be generally sharper, albeit by not a huge amount. Interesting…
There are of course other considerations to think about when choosing a lens such as distortion for example and it’s well documented that the 24-105L suffers a bit at the 24mm end. Although I am sure a 24L would do better in that regard, I know that by 35mm distortion is well under control, so on the this basis I would be hard pressed to choose the prime over the zoom. Sure you have an extra couple of stops for low light use that might come in useful for some, but don’t forget either that the zoom has built in image stabilisation which partially negates that difference, assuming you are not shooting a fast moving subject. A justification could be made for choosing either lens, but I have to say the zoom is doing extremely well. Time to look at F16 to see if anything significant changes.
35mm at F16
35mm prime @ F16 edge sharpness
24-105L on 35mm @ F16 edge sharpness
At F16 the edge sharpness between these two lenses is essentially identical. The prime might just have the edge by a hair, but others might disagree.
35mm prime @ F16 centre sharpness
24-105L zoom on 35mm @ F16 centre sharpness
Again it’s very difficult to separate the two lenses here at F16. I’d give this to the zoom but either lens would deliver perfectly usable results.
In summary, I know from practical experience that both lenses are capable of producing fine quality results in normal use, so the decision could largely come down to a question of preference in the way you like to work. Those who prefer a smaller less obtrusive lens with fast maximum aperture have an easy choice in selecting the prime, but on a pure quality score I’d personally favour the zoom given its superior rendering of detail at F4 and F8.
Some will say this proves little by comparing an L series zoom against non L series primes, but I think they are missing a point here. For the price of 2-3 primes you can buy the zoom and have it all in one, but if the zoom is comparable or better in quality it certainly makes for an interesting buying decision. It would be no huge surprise if the L series primes were another step up in quality, however even ignoring the question of cost for a moment, there are times when it just isn’t practical to shoot with a prime when you are working against a tight deadline for example.
I really don’t buy the argument that if a lens isn’t an L series it is automatically not worthwhile considering, since many non Canon lenses such as the 85mm F1.8 are of proven high quality and don’t need a letter L to justify their place in any photographer’s camera bag. Ultimately it is always a personal choice but I hope some of my thoughts here may be of interest to others. In a few days I shall do another comparison against the cheap standard 50mm F1.8 but don’t let the price put you off. That lens punches well above its weight.
For a long time I thought my Nikon 24-120 softness was my fault–user error, slow shutter speed–until I did some similar tests to this and found that the auto focus needed some fine tuning. I checked my other lens and adjusted all but one.
Is it possible that’s what is happening here?
As the previous poster said, I suspect focus could be an issue here.
I’ve never used the 24-105, but I have both the 35/2 and 35L and by F4 there is almost nothing between these 2 primes.
I would be surprised if the 24-105 was sharper than the 35/2 at F4 if the focus was accurate.
That’s the problem with most focus tests – when you see two photos A and B taken with the same camera, and A is sharper than B, you can say EITHER:
Lens A is sharper than lens B
OR
The focus for shot A was more accurate than shot B.
But just looking at the photos, you can’t tell which it is.
I’m not saying your results are wrong, just that I find it surprising given my experience with the prime, and that focus may be an issue here.
I wouldn’t exclude anything but I can say that a few years ago all my primes were checked out by Canon and found to be OK. As I pointed out in the blog post as well, the 35mm lens is one that I have used for years with good results and simply because it might look a tad softer here than the zoom when viewed in direct comparison doesn’t mean you couldn’t produce beautiful results with the prime.
It seems to me that the 35mm F2.0 is an old design going back at least 15 years and that it may simply be struggling to match the technology of a more modern L series lens.
I have deliberately refrained so far from looking at the images done with the 50mm, 85mm and 100mm macro lenses but my gut feeling says they will do much better and I have felt for some time now that they hold up to much greater scrutiny than the two wide angles.
Ever since the original 1Ds we have been hearing from photographers about their wide angle lenses not being good enough for full frame high resolution sensors and especially at the edges. I know of some who opted for alternatives from Carl Zeiss etc. with adaptors despite the inconvenience.
For the record, I have a Tamron 17-35 which I have used about twice in the last 4 years because the edge sharpness is a disaster until I stop down to at least F11 and preferably F16. Put that lens on my old 20D though and the results and brilliant.
I appreciate your tests so don’t take my comments the wrong way. I would believe the L zoom might be better in the corners than a non-L prime perhaps at F4, but not in the centre. I still think that if focus is correct the zoom would not beat the prime in the centre at equivalent apertures. I could be wrong, but my 35/2 is very close to my 35L in the centre once you get to F4. The corners are a different story.
It’s very tricky to do a correct lens sharpness comparison. Your target is not very contrasty – just a plain fence – easy for the AF to be a bit off. You should have a clearly defined dark target on a white background or vice versa. Are you just taking one or two shots using AF at each aperture? You should take several AF shots, resetting focus between each shot, as well as several liveview contrast detect focus shots, and some manual focus shots using 10x liveview as well. That’s about 10 or more for each aperture. Then take the best of each for each aperture. Sounds ridiculous, but you have to do it. Otherwise you might not be comparing lens sharpness, but comparing lens AF accuracy instead. Read the testing methods they use at http://www.slrgear.com for example. All manual and very precise.
Actually I meant take a look here:
http://www.slrgear.com/articles/focus/focus.htm
I can see a problem as you say you suspect that the further primes you wish to test may be better? Is it a good idea to approach a test with pre conceived ideas? of course not and is why on any particular day the same test may well give a different set of results. Sharpness is also a very crude way to judge lenses. CA , distortion contrast ect is for many a much more vital element and will make or break an image much more than whetehr it is sharp or not?
I make no claims jojo that this is the most thorough or scientific test in the world. Any supposedly advanced preconceptions I may have about the likely results from these lenses are simply based on the quality of results I have seen from past work. I won’t allow that to colour my analysis of the test images though when they are examined on screen and in any case readers will be looking at the same samples at 100% in the blog, so they can make up their own mind.
You are absolutely right that characteristics like CA and distortion need to be taken into account when viewing the overall qualities of a lens and indeed I have mentioned before that the 24-105L tends to suffer from more CA than my primes, but also that it has never been a problem to fix when Raw processing. I also mention in this test that the L series lens has a tendency towards higher contrast and warmer tones along with distortion being well under control by 35mm.
As also mentioned by others, I have found focus accuracy to be a big issue when comparing 100% crops from different lenses. It really can make a massive difference. What I normally do now is focus on a window edge in a brick house wall from an angle of about 30-40 degrees. At 100% I can then find the area of best focus of both lenses and compare detail in the bricks. Even if focus was in slightly different places you can just compare the sharpest regions from each lens. Of course this does not allow you to compare edge sharpness as the edges are out of focus, but I think it is a better test for centre sharpness.
BTW I also use this method to determine the optimum MA (micro adjustment) setting now that I have a camera with this facility. I just try different settings until the best focused region of the photo coincides as nearly as possible to the intended focus point. I find MA settings made in this way tend to give better results with subsequent real life subjects than the recommended test chart type methods.
Hi Ashley, thank you for taking the time to create these comparisons. They were very helpful.
From the comments, IMO, there ought to be less emphasis on whether lens 1 is better than 2 because in the real world, should we really care all that much? We’re talking about 12MP+ images here, and a 100% crop that isn’t as sharp as another 100% crop… To me, the comparisons showed that the 24-105 can produce sharp images, and with such a nice zoom range, the 24-105 should cover my needs. I believe it all boils down to whether a photographer has the creativity to produce something using the tools in his/her bag =).
Cheers!
http://www.flickr.com/groups/canon24-105/