Test Chronology – New Canon 16-35mm II lens:
4/30/07 – I’m out of town tomorrow and can’t do the UW test quite yet, but the 16-35 L II arrived this afternoon and I did a very quick test. Tripod mounted with focus on the part of the board in the center that said "Church". Center point AF fell right there, and it had enough contrast to pop focus. Note that the other two boards are positioned at the lower left and right, far from the central focus and about the hardest test of corner sharpness a wide angle will ever have. F-5.6 @ 1/125th second at ISO 100 for both shots.
Here is the 16-35 L II:
Here’s the 17-40. Not massive difference in angle of coverage, certainly not enough to justify trading lenses on that criteria alone.
Here’s a screen grab for the corner comparison. Both at "Actual Pixels", 100% enlargement in Photoshop, no sharpening, or any other adjustments for that matter. From JPG, quality "L".
Based on this I’d say that the 16-35 L II is greatly improved over the 17-40. Going back to Craig’s original assertion, the dome is not going to improve the performance of what a lens does in the air, I am optimistic indeed that I can do better than the 17-40 by using the 16-35 L II underwater. Having good glass to begin with is a very big part of the battle!
The performance of this lens in the first topside test certainly justifies the time spent dialing in the ideal port and port extension combination. 16-35 is on the left and 17-40 is on the right. You may not be able to tell much from this small JPG, but seen large on my 23" monitor the difference is significant.
Note also the contrast appears to be improved in the 16-35, the 17-40 looking a bit muddy in comparison. Shot is open shade, both 1/125th second at F-5.6.
My first 10 minutes of testing likely cost me $1500 for a new lens. Damn!
05/07/07 – I’m back in the studio now, thinking about getting in the water with the new 16-35 II and playing around with the housing. Interesting to note that the zoom gear for my 17-40, which is the same as one would use on the old 16-35, also fits the new 16-35 II.
I can’t speak for all housings, but at least for Seacam, the existing zoom gear works for the new lens.
05/10/07 – Did my pool tests today. Seacam housing, swivel 45 viewfinder (so I did not have to get my head wet during testing protocol and it would go faster), superdome and fisheye port, and the following port combinations with each port:
no extension, PVL20, 25, 30, 35, 40, 45, 50, 55, and 60
Lenses tested: 16-35mm II, 17-40, 15mm with Tamron 1.4 teleconverter
As with topside test, I provided an impossible test on the lens. I set one of the exposure slates in the background center, and that’s where the lens focused. I then put two more exposure slates at right and left upper corners, about 2 1/2 feet closer to the lens than the point of focus, and worked at only F-5.6 (aperture priority). Hard to imagine any lens being stellar in the corners under those conditions.
Why? That can happen on the reef. You have a central area of focus and reef rising on the periphery in the foreground. Had I chosen to shoot across a flat plane, the results would have been dramatically different. The lens would have performed far better focusing along a flat plane.
But, how often does that come up in normal UW imaging? Right now I only cared about what port extension would give the best results for the way I would likely shoot on a reef.
OK, here’s the quick and dirty on what is a pretty interesting comparison. First, here is the 16-35 with no port extension and the superdome. THIS IS WHY WE TEST. It is horrible. Note also that it is not as wide as the later tests with a properly matched port extension. That was actually a surprise to me. (All distances were the same).
16-35mm II, best results with Seacam Superdome and PVL 60. Note that there is distortion in the corners. That is to be expected, but note also that there is no vignetting like you’ll see on the 17-40 sample in the next post. The 16-35 II has much better color and contrast as well. The lens is very sharp in the center.
17-40, best results with Seacam Superdome and PVL55. Note that this trumps my previous testing where I have recommended either a PVL 30 or PVL35. This camera and lens have recently been sent back to Canon for precise calibration for a back-focusing problem. It is now very sharp with all topside tests, with all my lenses. I deduce this is the more accurate test, and the way I will shoot moving forward. Your personal cameras may differ and you should do your own tests.
You won’t see massive differences from these small thumbnail images. In reality, I did not see massive differences between the 100% enlarged view in the corners of the UW shots with the 17-40 and 16-35 II either, even in PS on a 23" monitor working from the RAW images. Here’s how I see it:
1. 16-35 II is definitely sharper than 17-40 topside. This is evident in the topside tests, both center and corners. The 16-35 II is significantly improved over the old 16-35 and 17-40.
2. 17-40 has issues with vignetting and is rather flat in terms of contrast and somewhat muted in terms of native color rendition (both easily tweaked in Photoshop).
3. 16-35 II is sharper in the center, even underwater.
4. If you think the 16-35 II is going to give substantially better corner resolution in a housing you’ll likely be disappointed. This is a dome issue, rather than a lens issue.
5. Is the 16-35 II worth it? For me yes. For topside resolution, for color and contrast purity, and for the slightly wider angle of view. But, don’t expect miracles in corner resolution when used with a housing. Future testing will have to look into how a diopter might affect performance.
6. Corner performance will be significantly enhanced by working at smaller apertures.
7. As expected, superdome was better for these lenses than the fisheye port.
8. Optimal resolution in the corners remains a dome issue, with refraction and virtual image making corner performance an ongoing challenge.
15mm and Tamron 1.4 teleconverter. Superdome and PVL 20. This is an interesting concept, suggested to me by Walt Stearns. I love the angle of view and comparative lack of distortion, but find it soft.
Final test was with no port extension, admittedly the worst combination for the 16-35 II and superdome, but at 2 different apertures. Look at the difference F-stop makes in terms of 100% view of the corner.
Here is F-5:
Here is F-9:
Clearly, aperture is critical in working with all of the lenses tested. I predict that at F-8 and smaller, the 16-35 II will do just fine, in the centers and in the corners, for most UW imaging applications. Real in-water testing, on a coral reef, needs to be the next step. But these tests reveal the start point.
As for the fisheye port and the 16-35mm II lens, the superdome was substantially better in terms of optical performance. I couldn’t find a port extension that really gave stellar results, but the better, but the best I could do with the fisheye port was with the PVL50.
As an aside, I haven’t found a Canon lens that is especially well suited for the fisheye port yet, but it is very good for the lens it was designed for, the Nikkor 16 (and now the 10.5mm for the cropped Nikon digital sensors). It is curious that while both the Canon and Nikkor fisheyes are nearly the same angle of coverage, they don’t perform identically in the same port.
For Canon I prefer the superdome for 15mm and 16-35mm II/17-40. For Nikon either fisheye port or superdome will work with both 16mm and 10.5mm.