Acceptable sharpness with wide-angles

"Sharpness is a tool, like colour, and, like colour isn't always necessary or (perhaps) even appropriate."

Whilst this is very true, my point is that the option should be available to produce sharp underwater images with rectilinear lenses. At the moment we can't - or certainly not to the same standard as is accepted on land - and this limits both the quality and the usefulness of ultra-wide (weitwinkel) underwater imagery. There may be a very good reason for a stunning image, which is not technically perfect, being perfectly acceptable, but there are many reasons for wanting technical perfection. At the moment it would be difficult to shoot multi-image panoramas underwater (useful perhaps for scientific reasons) as corners would make stitched pictures tricky. Fisheyes do deal with the problem to an extent but de-fishing them has the same effect as shooting on a lower MPixel camera.

I wonder how many underwater photographers would be interested in a port which overcame corner softness (Coustean apparently had domes made which were thinner at the edge than the centre as an example - though not necessarily a good one)?

I wonder how many underwater photographers would be interested in a port which overcame corner softness (Coustean apparently had domes made which were thinner at the edge than the centre as an example - though not necessarily a good one)?

I would definitely be interested since I'm shooting a Canon 14mm...

however, I would anticipate that I'll not be able to afford such a beast and since I'm shooting a 5D I'm learning to simply compose for cropping

I'm most disappointed with my WA close ups as the Nikonos 15mm was such a beauty for that. Even when there was no suitable WA (like lembeh straits) I was able to get some decent shots.

My current set up just doesn't do it for me the same way...

>>Nikonos rectilinear water contact primes had better corner sharpness than dome ports<<

The Nikonos lenses were categorically sharper at the edges, for good reason. As others pointed out, the water contact optics were designed from the start to be a simple w/a lens with a perfectly matched dome.

The reason no SLR/DLSR housing will ever match the Nikonos has to do with the virtual image created by the dome. The VA is effectively a curved image, and any land lens if optimized to focus on one which is flat. Larger, perfectly matched domes will minimize the problem, but it never goes away completely. The Nikonos dome, OTOH, never gets used in air so the internal optics can be matched to the curved image. Everyone lived with the soft corners for lots of reasons, like lens choices, reflex viewing, etc.

I was never that bothered by the soft corners, except in some rare cases; a little cropping can take care of it too. As for the general argument about sharpness being just "another tool", I would have to disagree, I think it's pretty damn important. Even good shots with motion blur have a kind of sharpness to them. FWIW, the stock agency I work with says specifically that they're not interested in images that aren't sharp. I suspect that some of Amos Nachoum's images seemed soft since they're of large animals in extreme environments, where you may accept some technical shortcomings. That said, I've seen of a lot of his work which is awesome in every respect...

Cheers...

I think that I am right in saying that the Nikonos 15mm had a field of view similar to that of a 20mm on 35mm FF. It was also a none retrofocus design. Meaning that it didn't have either the characteristics of an ultra-wide (weitwinkel) behind a dome in terms of coverage, nor was it designed to be scrutinised by todays unforgiving FF sensors (and being a fairly none retrofocus design it would probably not have worked very well with a digital sensor anyway).

The curved image is indeed the basic problem, but there have been attempts to deal with this (such as the Ivanoff corrector) and more modern optical designs do exist. As I see it, unless there is an attempt to produce a better soution that a simple dome, we will never be able to fully utilise the potential of digital sensors and ultra-wide (weitwinkel) lenses. Megapixels will no doubt increase, and the dynamic range of these sensors will too. But domes will be problematic until an answer is found to their inherent flaws.

I think there's general agreement that hardware solutions do exist for the dome problem, but that economics are the obstacle. No company is going invest huge R&D in an expensive solution so few u/w photographers would spring for; just look what happened with the Nikonos RS... I've already decided that if I win the lottery, I'll hire a bunch of hotshot optical engineers and have them design my ultimate vanity system exactly the way I want it. Then I'll start a foundation to give away the technology to those who deserve it... ;>)

My guess is the Nik 15mm would hold up pretty well on a FF sensor, if there was way to jury-rig one, but maybe there are other issues.

"My guess is the Nik 15mm would hold up pretty well on a FF sensor, if there was way to jury-rig one, but maybe there are other issues"

The sensors fitted to current dSLRs work best when struk by light as perpendicular to their surfaces as possible - which means that wide-angle lenses need to be of retrofocus design (in theory the latest Nikon lenses are - but there are no ultrawide (weitwinkel) fixed primes of say 12mm focal length so users are stuck with slower zooms so far). The Nikonos 15mm (even the later version) was not a substantially retrofocus design so the light paths would have neen very oblique in the corners - hence why it would probably not be a good lens for digital.

Leica's new M8 uses a sensor designed to work better with more obliqe light and if someone was enthusiastic enough it might be possible to use it with the Nikonos 15mm - that would be a very interesting idea BUT with a 1.3x sensor crop..... back to square one!

If we are lucky in the future sensors might be better designed for wide-angle lens work and have a higher dynamic range - rather than simply more pixels. But they cannot overcome the inherent problems with using simple domes underwater.

how about a ultra wide (weitwinkel) lens (maybe the 15mmUW-Nikkor) on a FF sensor array that is laid down on the inside of a sphere rather than a flat surface as they are now. Sure it might only be good for the one wide lens, but hey, it might be sharp!

Sounds like we're getting to the heart of the matter here. The Nikonos 15mm is a perfectly matched system where the main optic is designed to handle a curved virtual image from the dome. The only SLR lenses we have available that do this are the fisheyes.

You'd have a job fitting the 15mm Nikonos to an SLR, because of the short retrofocal distance. Bonk, bonk! There's the mirror. Same with most wideangle rangefinder lenses. See how the rear element sticks out:

Yes, the 15mm is the same FOV as a 20mm housed topside lens, 94 degrees. By 20mm on a FF SLR we're getting back into sharp corner territory. Of course the upsides of the housed lens include the ability to do split level shots and being attached to a DSLR, providing immediate feedback, more than 36 shots, autofocus and all that good stuff.

Leica have done valuable work with their offset microlenses, but of course they had a bigger problem than the DSLR guys to start off with, namely a more extreme angle of incidence at the edge of the frame. Vignetting is largely a none issue for a FF DSLR like the 5D. If it deemed undesirable, it can be removed very easily at the RAW conversion stage.

Martyn