18 replies to this topic

[StephenFrink]

I just spent an hour doing pool tests with Mike Mesgleski. He brought his Seacam D200 housing with both 10.5 Nikkor and Tokina 10-17. We tested at all apertures from F-4 through F11 with both lenses, tested with Seacam Fisheye port and Superdome, and with PVL20 port extension and without.

I'll spare you (and me) all the thumbnail analysis, but am willing to send a CD to anyone anal enough to want to look at all these pictures at 100% maganification. Here's the synopsis of important comparisons:

1. Fisheye port and Superdome with 10.5 Nikkor - corners are pretty bad until F-8. Superdome is a tiny bit wider in coverage, Fisheye port ever-so-slightly sharper. Not much difference in the port performance, but still surprisingly bad at apertures wider than F-8. Tried both ports with PVL20 (the smallest Seacam makes) but it vignettes badly.

2. Tokina 10-17 versus Nikkor 10.5 - Tokina is significantly sharper at wider apertures but unfortunately also exhibits very significant color fringing. See this the 100% crop, shot a 1/2000 second so what you're seeing is not motion but fringing. The 200% crop really reveals the issue. Look at the black vertical line that goes blue, and the blue halo at the right side of the white edge of the slate.

100%


200%


The lens is apparently pretty sharp, and quite versatile, but there's no free lunch. It has other optical flaws that may or may not be revealed in most underwater scenes.

3. Ideal port configuration for Tokina 10-17 - There is little difference between the fisheye port and the fisheye port + PVL20. Both are pretty good, but there is a little vignetting from dome shade with PVL20. Use Fisheye port with no port extension. The Superdome + PVL20 is better than the Superdome alone, and is easily the best of all port configurations tested for Tokina 10-17. However, there will be a bit of vignetting from the port sunshade. To use PVL20 you'd have to trim the shade a little. Tokina 10-17, superdome, no port extension is quite nice; and probably the easiest/best solution for all but the most anal pixel-peepers. Those will want to modify their sunshade and go with the PVL20.

There has been some discussion that the Tokina 10-17 was good enough for cropped sensor cameras that it might drive shooters away from full frame to the cropped sensor cameras for which this lens was designed. My test did not lead me to this conclusion.

Thanks to Mike Mesgleski for contributing the hardware for the test, and for actually taking the shots. I was just the slate-holder.

[Arnon_Ayal]

Thanks for the useful info.
If it possible I'll be happy to see side by side 10.5 Vs 10-17.

[StephenFrink]

Here is the full frame view at both 10mm and 17mm. The lens is extremely useful in terms of angle of coverage and convenience. If you can live with the fringing it is an excellent option.


10mm


17mm

[StephenFrink]

Thanks for the useful info.
If it possible I'll be happy to see side by side 10.5 Vs 10-17.

Does this help? Tokina is on the left:

[MikeVeitch]

aha! finally a good UW direct comparison of 10 and 17.

Thanks Steve and Mike.

[Lionfish43]

I guess it depends on your shooting style but I rarely, if ever, shoot uw photos at f4. You're presenting the worst case scenario.

[dysert]

I just spent an hour doing pool tests with Mike Mesgleski. He brought his Seacam D200 housing with both 10.5 Nikkor and Tokina 10-17. We tested at all apertures from F-4 through F11 with both lenses, tested with Seacam Fisheye port and Superdome, and with PVL20 port extension and without.

I'll spare you (and me) all the thumbnail analysis, but am willing to send a CD to anyone anal enough to want to look at all these pictures at 100% maganification. Here's the synopsis of important comparisons:

1. Fisheye port and Superdome with 10.5 Nikkor - corners are pretty bad until F-8. Superdome is a tiny bit wider in coverage, Fisheye port ever-so-slightly sharper. Not much difference in the port performance, but still surprisingly bad at apertures wider than F-8. Tried both ports with PVL20 (the smallest Seacam makes) but it vignettes badly.

2. Tokina 10-17 versus Nikkor 10.5 - Tokina is significantly sharper at wider apertures but unfortunately also exhibits very significant color fringing. See this the 100% crop, shot a 1/2000 second so what you're seeing is not motion but fringing. The 200% crop really reveals the issue. Look at the black vertical line that goes blue, and the blue halo at the right side of the white edge of the slate.

100%


200%


The lens is apparently pretty sharp, and quite versatile, but there's no free lunch. It has other optical flaws that may or may not be revealed in most underwater scenes.

3. Ideal port configuration for Tokina 10-17 - There is little difference between the fisheye port and the fisheye port + PVL20. Both are pretty good, but there is a little vignetting from dome shade with PVL20. Use Fisheye port with no port extension. The Superdome + PVL20 is better than the Superdome alone, and is easily the best of all port configurations tested for Tokina 10-17.

There has been some discussion that the Tokina 10-17 was good enough for cropped sensor cameras that it might drive shooters away from full frame to the cropped sensor cameras for which this lens was designed. My test did not lead me to this conclusion.

Thanks to Mike Mesgleski for contributing the hardware for the test, and for actually taking the shots. I was just the slate-holder.

What zoom gear works for the Tokina?
Pete

[StephenFrink]

I guess it depends on your shooting style but I rarely, if ever, shoot uw photos at f4. You're presenting the worst case scenario.

Does F-8 tell you what you need to know?

100%


In the 200% view you'll see the fringing is essentially the same. The resolution in the corners is certainly better at F-8, as would be expected. If you know the worst case scenario, you can always interpolate to enhanced performance, but know what to expect when you have to shoot wide open. My opinion anyway.

BTW, as I said, other than fringing, this seems a wonderful lens. Excellent performance in the corners, and when used with the right port and port extension far better than the Nikkor 10.5mm. Except for the fringing. And even that probably won't show up in all subjects and all colors. Something to look for in your own real-life images on the reef, perhaps.

[StephenFrink]

F-8, 200% view

[StephenFrink]

Final thought on fringing/chromatic aberration issue: I had pretty nice results from going to the Lens tab in the RAW protocol box in CS2 and pulling -45 on the red/cyan slider and +70 on the blue/yellow slider. It seems the issue is largely correctable, certainly easier than a lens that is lousy in the corners to begin with.

[MikeVeitch]

Excellent performance in the corners, and when used with the right port and port extension far better than the Nikkor 10.5mm.

I think you just upped Tokinas sales by quite a bit in the UW photo circles with that statement...

[Johnny Christensen]

I guess it depends on your shooting style but I rarely, if ever, shoot uw photos at f4. You're presenting the worst case scenario.

Hmmnn... f4 and 1/10sec at 200iso is a good day when I'm shooting. So that test has a lot of meaning for me.

[John Bantin]

Some domes are better than others too. This colour fringing is most revealed with divers in black wetsuits at the edge of frame. However it's best to keep divers away from frame peripheries with fish-eye lenses anyway! Chromatic correction in CS/2/3 helps.

[StephenFrink]

As a final observation I offer Mike Mesgleski's comments to me in an e-mail last night. He had the same RAW files on his computer and after analyzing the files came up with:


"I did a Tokina and Nikon side by side test for both the FE & SD.

With the FE (no PVL) Tokina beat Nikon at ƒ4 & ƒ5.6. They both tied at ƒ8 & ƒ11.

With the SD (PVL 20 w/ Tokina & no PVL w/ Nikon) Tokina beat Nikon at ƒ4 & barely beat Nikon at ƒ5.6. They both tied again at ƒ8 & ƒ11.

I see what you mean about the chromic aberration on the edges. True, not necessarily a deal breaker, especially if it is correctable when noticeable."


It is interesting that with access to all the RAW data, he came up with the same conclusions. Also interesting that for someone who already has the 10.5mm Nikkor it will perform on par with Tokina at F-8 and smaller apertures. it won't have the versatilty of the zoom lens of course, but is still a fine wide angle tool with excellent close-focus capability, especially stopped down a little.

[Walt Stearns]

Coming from the full frame camp, Tokina’s 10-17 fisheye zoom is a very compelling lens.

Just finish playing with one for a week out on the Texas Flower Gardens. Have to admit the lens is not only pretty sharp, but quite versatile. As Steve did mention there's no free lunch. Getting good corner to corner performance from wide angles once you step away from fisheye’s (10.5 with crop or 15’s with full frame cameras) can be a real bitch.

In my own dealings with the Canon 5D, the 15mm fisheye works fine, the same lens with the 1.4 Tamron TC/ 2Omm extension ring with the Subal FE2 dome port is a pretty slick, but not perfect alternative for something less wide. Don’t know where I would have been otherwise without it when I shot the goliath grouper aggregation shots in the resent issue of the Underwater Journal.

As for the Canon 17-40 or 16-35, I think we are still looking for the right port/extension ring/ diopter combo to make them work as well underwater as they do above.

Optical flaws with the 10-17 I think we are splitting hairs here. Yes, even at 17mm, anything with a straight line won’t appear straight. It is after all a fisheye, although it is also a wide-angle zoom. For me, seeing even slight curvature in underwater scenes is largely a non-issue. Seeing corners turn to mush is.

Now have to agree fringing/chromatic aberration could be an issue. Then again is it when most of it can be largely corrected by going to the Lens tab in the RAW protocol box in CS2.

In this sample taken on a reef 70 feet down, I pulled the red/cyan slider to –35, with the blue/yellow slider moved up to +71 on.

In the end, it is certainly easier issue to deal with than lousy corner sharpness, or complete lack of.

Results may vary with port and environment locations, but right now, my vote is on this lens.

[TomStack]

Stephen,

All I can say is a large "Thank You!" to both you and Mike for taking the time to conduct these tests!

Been debating this for some time, but no doubt that the Tokina is my next lens purchase.

Best,

Tom Stack

http://www.tomstackphoto.com

[AllisonFinch]

Glad to see this. I just received my Tokina 10-17 for my Canon. Can't wait to put it through it's paces. I just need verification of it going on my insurance policy. It was a pricey little bugger!

[StephenFrink]

Update: As per correspondence with Andre Crone, now on location shooting with Tokina 10-17 and testing PVL20 and Superdome, he is seeing slight vignetting from sunshade, primarily in upper left corner. My comments to him after reviewing his image and our test shot:

"Thanks so much for sending your JPG.  Now that I see what you are talking about it all becomes clear.

First, when I saw that the vignetting is exactly in the corner, that tells me that the click stop on the port and port extension is fine.  What you are seeing is consistent with a port shade showing up in a wide angle image.

So, now having seen that I went back to my pool of test images and found exactly the same vignette spot in our test upper left, exactly where it is most pronounced in your sample.  I am very surprised to not have noticed previously. Mike and I were so obsessed with studying corner resolution and all things happening on the type of the exposure slate on the right side of the frame we did not notice the upper left apparently.

Here's the dilemma.  The PVL 20 and superdome does give the very best corner performance. Superdome alone is good, but clearly not quite as good.  There can be no smaller PVL than the 20 because you need threads to screw into the housing on one end (male), and the ability to screw the port on at the other end (female).  PVL20 is as small as the port extension can be machined.

So, here are your options:

1.  shoot with pvl20 and either crop or clone the offending corners in Photoshop.  

2.  shoot with superdome only and accept a little softer corner.

3.  slightly file the corner that shows up from the dome port.

The Tokina 10-17 is a different beast than the 10.5 or any other fisheyes that have come before.  If it were my lens and housing I would probably go with superdome, PVL 20, and try to slightly file that upper left corner on the port shade; but that is an issue for you to decide."

[PRC]

Looks like it is the bottom left also.