16-35 mm F4.0 IS first impressions - corner sharpness

[joel_p]

<p>Used my new 16-35mm F4.0 IS for the first time today. Have to say I like it. Just in terms of sharpness, it is much improved over the 17-40mm in the corners (i can't compare to 16-35 2.8). Not much between them in the center. Just to give you an impression, I have posted a few shots taken at F4.0, tripod mounted, at similar speeds, 28mm. Taken at different times of day, so comparing a bit of apples/oranges, but you should get an idea. Cheers, Joel</p>

<p> </p><div></div>

[joel_p]

<p>2nd photo</p><div></div>

[joel_p]

<p>3rd photo - far corner</p><div></div>

[joel_p]

<p>4th photo - far corner</p><div></div>

[lumusphoto]

<p>What was it like at higher apertures? I recently reviewed my 17-40L <a href="

and once stopped down the corner detail was very impressive. There are test shots in the video. Am interested to see more testing from the 16-35 as it looks like another cracker from canon!</p>

[colin carron]

<p>very interesting. Leon is right about stopping down the 17-40 but I have still been checking the ebay completed items for an idea of what my 17-40 might fetch!</p>

[joel_p]

<p>I'll leave the comprehensive testing to the pros, but here are the center, corner shots at 5.6 (which was the sharpest of 4.0, 5.6, 8.0 and 11.0 apertures).</p><div></div>

[joel_p]

<p>corner at 5.6</p><div></div>

[joel_p]

<p>here is the entire shot</p><div></div>

[lumusphoto]

<p>I'm looking forward to testing this for my youtube channel. It definitely looks an improvement on the 17-40L but saying that it is 11 years old now lol! Thanks for posting the images though :)</p>

[dcstep]

<p>You should consider looking at results after Raw conversion with Digital Photo Professional and applying Digital Lens Optimization in Raw conversion. If you're using these lenses and not applying DLO, you're not getting the most out of your lenses.</p>

<p>People still test lenses as if they're still using film. With digital sensors, you have filtering and correction during the capture process, such that the resulting image doesn't not necessarily represent the actual ability of the lens. Applying DLO makes correction at every focal length and every aperture to correct for distortions in the lens and introduced by the camera during the capture process.</p>

[Robin Smith]

<p>Distortion correction will be destructive at the pixel level as one can't stretch, eliminate, or duplicate pixels without losing detail: if one worries about such things.</p>

[dcstep]

<p ><a name="00cgSi"></a><a href="/photodb/user?user_id=23754">Robin Smith</a> said:</p>

 

<blockquote>

<p>Distortion correction will be destructive at the pixel level as one can't stretch, eliminate, or duplicate pixels without losing detail: if one worries about such things.</p>

</blockquote>

<p>This line of thinking assumes that the filtering and digital capture process doesn't do any damage, as if we were shooting on film. Use the same lens on three different body models from the same manufacturer and you'll get three different results. That's what we're talking about correcting.</p>

 

[g dan mitchell]

<p>"Distortion correction will be destructive at the pixel level as one can't stretch, eliminate, or duplicate pixels without losing detail: if one worries about such things."<br>

<br>

Well, yes, but...<br>

<br>

It turns out that the potential for loss of detail is often greatly overestimated by those who haven't done a lot of such correction in post. If the corrections are not of a huge magnitude, you usually cannot see the difference even in very large prints.<br>

<br>

I posted a few examples at my blog:</p>

<ul>

<li><a href="http://www.gdanmitchell.com/2007/09/16/a-test-correcting-lens-distortion-in-post-processing">Correcting Lens Distortion in Post-Processing</a></li>

<li><a href="http://www.gdanmitchell.com/2009/02/20/a-test-correcting-perspective-in-post-processing">Correcting Perspective in Post-Processing</a></li>

</ul>

<p>Dan</p>

[dcstep]

<p>Excellent post and examples Dan.</p>

<p>DOL have been improving steadily since Dan's original post.</p>

<p>It's easy enough to try for youself. DLO comes as part of DPP, which ships with any Canon DSLR.</p>

[Wouter Willemse]

<p>Still, those lens correction technologies - which are not part of the lens - should you really be testing a lens with those enabled to see what the lens is capable of? It's seriously a question, but obviously I have my doubts.<br /> The capabilities of the camera and software should be excluded or normalised as much as possible when testing a lens. If you want to test a total system with the lens as an integral part of a chain, it's a different thing. But if the lens has chromatic aberrations, distortion or whatever, lens testing should reveal it - not hide with software optimisations. Or am I completely missing a point now?</p>

[dcstep]

<p ><a name="00cgYM"></a><a href="/photodb/user?user_id=5189561">Wouter Willemse</a> <a href="/member-status-icons">said:</a></p>

 

<blockquote>

<p>Still, those lens correction technologies - which are not part of the lens - should you really be testing a lens with those enabled to see what the lens is capable of? It's seriously a question, but obviously I have my doubts.<br /> The capabilities of the camera and software should be excluded or normalised as much as possible when testing a lens. If you want to test a total system with the lens as an integral part of a chain, it's a different thing. But if the lens has chromatic aberrations, distortion or whatever, lens testing should reveal it - not hide with software optimisations. Or am I completely missing a point now?</p>

</blockquote>

<p>Good point Wouter, but I've thought about it a lot also and come to the opposite conclusion, because we can't escape the fact that the body digitally alters the image during capture. Still, I respect your position. If we could test a lens without using a digital body, then I'd agree with you. Unfortunately, the lens is part of a digital system; body/lens/software.</p>

<p>Of course, I realize that testing labs face a huge problem because we all use so many software. I suppose that if it were a Canon lens on a Canon body, then they could show the results from Digital Photo Professional. That would be meaningful and as the manufacturer intended. I think that all the major manufacturers have their own Raw conversion software, but I don't know if all have DLO, as Canon does.</p>

<p>Unfortunately, DPP does not offer DLO for non-Canon lenses, like Sigma and Tamron. For those, the tester could use DXO, LR or another Raw converter, but then the test becomes as much of a test of the software as it is a camera/lens combination. That's a big ole can of worms, BUT I'd still like to see some reviewers try to address it. They're really ignored it for the last few years. I haven't found an in-depth discussion of DLO in any of the major photography magazines.</p>

<p>For us regular users, I say do your comparisons with and without DLO engaged and decide for yourself. In my case, back in 2009, I was ready to send my EF 24-105mm f/4L IS back to Canon of a look see since I was having problems with softness and CA. When I started using DxO Optics Pro, my problems went away. The 24-105mm is now one of my go-to lenses.</p>

 

[g dan mitchell]

<blockquote>

<p>"Still, those lens correction technologies - which are not part of the lens - should you really be testing a lens with those enabled to see what the lens is capable of? It's seriously a question, but obviously I have my doubts.<br />The capabilities of the camera and software should be excluded or normalised as much as possible when testing a lens. If you want to test a total system with the lens as an integral part of a chain, it's a different thing. But if the lens has chromatic aberrations, distortion or whatever, lens testing should reveal it - not hide with software optimisations. Or am I completely missing a point now?"</p>

</blockquote>

<p>I'm of two minds about this. What we are really interested in is what level of photographic quality we can produce using the equipment — not the specific measured performance characteristics of the lens per se. When we use the equipment to make photographs we don't do so with one arm held behind our backs. We always use whatever techniques and features are available to us to optimize the image.<br>

<br>

Imagine that you had two lenses. One was natively quite good — good resolution, good control of various sorts of distortion, and so forth. The other was natively almost as good but not quite, but when used on your camera its software compensated for the slightly lower performance by correcting various sorts of distortion and, after doing so, the image quality from this second lens was better than that from the first.<br>

<br>

Which is better? Which would you rather use to make your photographs?<br>

<br>

While I see the objective value in understand the underlying uncorrected lens capabilities, I've often thought that rather than trying to isolate everything that a more interesting and useful test would be to put two pieces of gear into the hands of a very skillful photographer (or photographers) and see what the end result would be when everything is optimized in precisely the way that skillful photographers will optimize them.<br>

<br>

Dan</p>

[Wouter Willemse]

<p>Dan, Stephen, no disagreeing with what you're saying at all. It was an honest question; I am mostly undecided on the subject, also because to a very large extend, lens tests are rather meaningless to me - it's the photos that count. So it's the system that counts in the context of the photographer, style of photography and desired images... I use a number of lenses with noticeable optical flaws which to me render character to the image. They'd do terrible in tests - but in the real world with 3 dimensions, they deliver the goods. I prefer them over my lenses that do quite well in tests often.<br>

That said, I appreciate especially distortion correction when done right, and it's a useful tool which helps getting the best of things. And yet, after spending over €1000 for a lens (competitor lens to the 24-105L with a different mount ;-), to see it has visible distortion prior to correcting it, is a pity. No issue in the real world, but it's a lot of money we're spending, and there is a certain feel-good factor in knowing it performs well without the help of additional tools.</p>

[dcstep]

<p>Understood Wouter.</p>

<p>I agree that the proof is in the shooting. For zoom lenses, like the OP 15-35mm, shot at a wide variety of apertures and focal lengths, I can't imagine a user preferring the results uncorrected vs. using a competent DLO program.</p>

<p>The expense of high quality lenses is indeed quite high, as you state, so I like to maximize my chances for a sharp image by engaging DLO. BTW, other than fisheye images that I often de-fish partially, I've never backed off the DLO to get a better result. That includes a number of 50" prints.</p>

[dcstep]

<p>Dan said:</p>

<blockquote>

<p>While I see the objective value in understand the underlying uncorrected lens capabilities, I've often thought that rather than trying to isolate everything that a more interesting and useful test would be to put two pieces of gear into the hands of a very skillful photographer (or photographers) and see what the end result would be when everything is optimized in precisely the way that skillful photographers will optimize them.</p>

</blockquote>

<p>I'm of this mind and I think Wouter is also; however, I've never really seen this happen.</p>

<p> I also avail myself of the luxury of borrowing a lens and/or body from Canon Professional Services to see how the images look on my computer. People that don't belong to CPS can rent a lens for a short period, or rent two when you're narrowed down that far.</p>

[lumusphoto]

<p>Most people do not go to this anal level of testing - even us shooting everyday for a living. We want real world results as time is money to me whole point of the progress of a lens such as the 16-35 f4 is to spend less time in front of a computer and to maximize results in the field. I understand the need for this level of testing but there also needs to be some real world results for people using this equipment day in day out and relying on it to produce the result which will inevitably put food on the table.<br>

I do however appreciate what you guys are saying and understand the need for vigorous robust testing.</p>

[dcstep]

<p>Remember, you can't "maximize the result in the field" if your shooting digital. The capture process introduces errors and the Raw conversion process can correct errors and/or add errors. Luckily, applying DLO correction doesn't add to Raw conversion processing time. You either turn it on and leave it on, or turn it off. The processing is automatic and requires no time.</p>

<p>When we shot Kodachrome, Ecktachrome, Fujichrome, etc., we tried to get exposure perfect. If you didn't "maximize the result" in the camera, there wasn't much to save or do. Now, when conditions allow, many (most?) of us "expose to the right", over exposing the image so that the uncorrected image looks washed out. This practices maximizes the color and brightness information in the Raw file. In Raw conversion, we then "normalize" the exposure and contrast to get it to the image that was in our mind's eye. It's a totally different world.</p>

[g dan mitchell]

<blockquote>

<p>"Remember, you can't "maximize the result in the field" if your shooting digital. The capture process introduces errors and the Raw conversion process can correct errors and/or add errors."</p>

</blockquote>

<p>The point you are trying to make here is unclear to me.<br>

<br>

dan</p>

[Marcus Ian]

<blockquote>

<p>Still, those lens correction technologies - which are not part of the lens - should you really be testing a lens with those enabled to see what the lens is capable of?</p>

</blockquote>

<p>No. No you should not. Simply put, while a lens is simply a tool to help us achieve imagery, there is no way to compare the optical quality of lenses (especially from 3rd party manufacturers) to each other with out disabling the gimmickry. - Either that or simply shoot on film bodies - I understand there are still a few of those around...</p>

<p>While I would agree that while shooting it is probably best to choose to correct, when deciding on a lens, you should have all the available information - which includes what the OOC RAWS look like - how bad it<em> really</em> is. That way you can make a direct comparison between one lens and another.</p>

<p> </p>