39mp underwater

Wow, I'm impressed. What crop if any? Looks a little over sharpened though.

Hey Loftus,

A little sharprning was used. The crop resulted in aprox 7x10 @ 300dpi. Again, I don't think my set up is for super macro but, the results are good enough I think. I will play around with settings to determine the sharpest results next weekend. I think it can get better.

Troy

New photos

Hey, Troy. I heard last year at DEMA that Patima (Korean housing company) was planning on selling digital MF housings made out of titanium. I don't know if they ended up pursuing that project, but it would be pretty cool.

As far as I know, Patima was looking for at least 4 buyers to start the project & he did not get the order. They will start selling D300 housing soon.

Sam

Hi Troy,

Thanks for the new pics. I think the first is the best. Having seen a few images now and giving this what I think is a fair assessment, I have to conclude that the constraints placed on this camera by port / lens combinations significantly limit the capabilities of this camera, to the extent that high end full frame and cropped sensor cameras are every bit as good as this camera underwater, in almost any situation, and actually better in some situations. Similar issues with corner sharpness etc are seen with this camera behind a dome and it may be worthwhile to do some experimentation with diopters to see if this can be improved. If not I would have to say that given the option of a full frame Canon 1DSMk111 or this camera, underwater, I would choose the former, at any price.

Good morning Loftus,

It has been awhile as I have been very busy. Thanks for the comments. I agree, in fact the dome port that was given to me by UK-Germany is an 8" dome port. The 28mm is huge. I don't understand why they did this. As I understand it now, there is a Super Dome port (9.5 or 10in) offered by Subal or Seacam. With some modifications would this not have been a better option? The corner to corner sharpness is perfect out of water and the software (Flexcolor) corrects any other related issues. Is there any info someone can send my way to get a better dome. By the way, I have the 1dsmk3 and I am very happy with it. Not ready to take it u/w though.

Thanks,

troy

Just got off the phone with Uwe at Uk-Germany and was told that they are building an acrylic 1meter cube pool to physically test for port corrections to correct the corner to corner issues. His estimate is around 6 weeks to have good results. This is good news for me and I will post more on the results with photos.

Thanks

troy

Good morning Loftus,

 

It has been awhile as I have been very busy. Thanks for the comments. I agree, in fact the dome port that was given to me by UK-Germany is an 8" dome port. The 28mm is huge. I don't understand why they did this. As I understand it now, there is a Super Dome port (9.5 or 10in) offered by Subal or Seacam. With some modifications would this not have been a better option? The corner to corner sharpness is perfect out of water and the software (Flexcolor) corrects any other related issues. Is there any info someone can send my way to get a better dome. By the way, I have the 1dsmk3 and I am very happy with it. Not ready to take it u/w though.

 

Thanks,

troy

I don't think there's a substitute for in water testing, so hopefully UK will work something out. I don't think you can be guaranteed better performance just because the dome is larger so probably you should see some tests, maybe with different diopters before buying anything. You may also want to contact Ryan at Reef photo, he is working on some custom domes.

I was going to mention Ryan as well. He has a large dome in development and should be able to brainstorm an adapter.

Thanks guys,

According to Uwe, the size of the dome isn,t the problem. Rather it's the distance from glass to port. Since my 22mm equivilant w/a lens is not super wide maybe this is true, but the lens is huge. I don't know if this plays a factor in corner blur. I will contact Ryan at Reef. Thanks for the lead.

Troy

Troy

Your problem is that you are using a simple, single element lens system (ie a concentric dome) in front of your sophisticated wide-angle lens. The use of a 'simple' lens system like this is bound to be the limiting factor.

Without wanting to go right into the 'basic' theory here, you will need to correctly position your lens in the dome as stated BUT given your high MPixel system and the probable lens quality, I'd suspect that image field curvature created by the dome will always be a problem. All the problems faced by smaller format dSLRs than yours will be magnified by a more exacting high resolution, larger format so you will probably find that repositioning your lens and port positions will help deal with some but not all problems.

There have been various methods tried to overcome wide-angle problems but in general it is accepted that beyond about a 90 degree field of view, image curvature is a limit on rectilinear wide-angle designs being used behind concentric dome ports. You are approaching this limit with an exacting system so you may well be optically constrained earlier (ie at lesser wide angles). The only solution is to use fisheye lenses or get someone to design a non-concentric dome port specifically for your lens (could be an exceedingly expensive option). I've also wondered if perhaps an aspheric diopter might be a solution but found nobody able to supply anything (something like a varifocal spectacle lens?!).

Uwe is buliding a small testing pool. So lets see what he comes up with. Any improvemnet in corner sharpness will be greately appreciated.

troy

I spoke with a housing designer about housing a 40mm Distagon (retrofocus lens design) Hasselblad lens a number of years ago. This lens has an 88 degree (diagonal) angle of view on the 2 1/4 inch square format. My recollection is that he said it would require a 10 inch dome port. The Superwides (38mm non-retrofocus design, 90 degrees) did OK with 8 inch dome ports.

Dome size (diameter) and lens field of view/focal length are NOT fundamentally related!

The dome diameter dictates the distance of the curved virtual image it creates from the dome.

The idea of a diopter is to allow a lens to focus on the centre of the curved virtual image (of a subject at infinity) when the lens in use is physically adjusted so that it is also focused at infinity - ie it retains the full focus of the lens in use. The diopter used is in this first instance dictated by the dome diameter, although strength may be varied depending on the close focus ability of the lens in use. Some lenses may not even appear to need a diopter as they can focus close enough to focus on the virtual image, but for other optical reasons they may be beneficial.

The problem with subjects which have depth to them (ie close corner detail) is that the close corner subject matter may lie at the corner of a cuved virtual image very close to the port indeed (from the camera's viewpoint) and this point lies so close to the lens that it is not possible to retain its 'virtual' position in focus by stopping down. That is to say, there is insufficient depth of field to enable the required 'thickness' of the curved virtual imagery to be all in focus. This is the basic problem with using concentric dome ports!

Solutions include accurate alignment of dome port and lens, use of appropriate diopter, using as small an aperture as possible and last but not least, choice of subject matter which exhibits the problem least (ie minimise close corner detail).

However the limit for which 'good' correction is possible is accepted as being a 90 degree field of view, and with more demanding systems (such as digital MF or hi MP dSLRs) this is likely to reduce.

Sorry to labour this point, but you need to have a good understanding of the optical problems associated with dome ports when aligning/optimising lens/diopter/port combinations. In general, a larger port backs off the virtual image which is often helpful but limits close focus ability (where abberations due to using a simple lens system, will kick in as you get too close). Dome ports are a compromise!

Tom on your comment, the Biogon 38mm is now a very old lens indeed (a 50s design I believe) and whilst it was housed, I suspect that in film days the resulting images were adequate rather than startling. Troy's system is likely to be far more demanding.

Dome size (diameter) and lens field of view/focal length are NOT fundamentally related!

 

The dome diameter dictates the distance of the curved virtual image it creates from the dome.

 

The idea of a diopter is to allow a lens to focus on the centre of the curved virtual image (of a subject at infinity) when the lens in use is physically adjusted so that it is also focused at infinity - ie it retains the full focus of the lens in use. The diopter used is in this first instance dictated by the dome diameter, although strength may be varied depending on the close focus ability of the lens in use. Some lenses may not even appear to need a diopter as they can focus close enough to focus on the virtual image, but for other optical reasons they may be beneficial.

 

The problem with subjects which have depth to them (ie close corner detail) is that the close corner subject matter may lie at the corner of a cuved virtual image very close to the port indeed (from the camera's viewpoint) and this point lies so close to the lens that it is not possible to retain its 'virtual' position in focus by stopping down. That is to say, there is insufficient depth of field to enable the required 'thickness' of the curved virtual imagery to be all in focus. This is the basic problem with using concentric dome ports!

 

Solutions include accurate alignment of dome port and lens, use of appropriate diopter, using as small an aperture as possible and last but not least, choice of subject matter which exhibits the problem least (ie minimise close corner detail).

 

However the limit for which 'good' correction is possible is accepted as being a 90 degree field of view, and with more demanding systems (such as digital MF or hi MP dSLRs) this is likely to reduce.

 

Sorry to labour this point, but you need to have a good understanding of the optical problems associated with dome ports when aligning/optimising lens/diopter/port combinations. In general, a larger port backs off the virtual image which is often helpful but limits close focus ability (where abberations due to using a simple lens system, will kick in as you get too close). Dome ports are a compromise!

 

Tom on your comment, the Biogon 38mm is now a very old lens indeed (a 50s design I believe) and whilst it was housed, I suspect that in film days the resulting images were adequate rather than startling. Troy's system is likely to be far more demanding.

Although more than 50 years old, the Biogon lens is one of the best wide-angles ever designed. It is unrivaled by most current lenses. The main improvement was the application of multicoating, which has only been around since about the 70's. There are Biogons still being made including some of the sharpest lenses in Leica M mount. The current 21mm f/4.5 has a design that is very similar to the 38. One reason for the 38’s reputation is that it is a non-retrofocus design. This may have influenced being able to adapt it to UW use, see my last question.

The point of dome diameter: A dome of a given diameter will project a virtual image that is independent of the camera lens, etc. However, I assume a larger diameter done will project a virtual image that is has a 'larger radius of curvature' than a smaller one. For example, when a given lens is focused on the middle part of the virtual image (on the optical axis, the very center that is parallel with the plane of focus), the periphery is focused further away than the center. This is easily seen in some fisheye shots: focused on a subject in the middle and the background near the edge of the frame is in focus, but background in the center of the frame is not. How does the shift to farther focus within the field (as angle from the optical axis) change with dome size? What is poorly described in the literature is the shape of virtual image – it is only spherical at certain distances, aspherical at others. That is why I put larger radius of curvature in parentheses above.

BTW, Does the position (does the size appear larger or smaller) of the entrance pupil of a retrofocus lens change with angle as measured from the optical axis? Is this at the core of the issue with this Troy’s lens? Symmetrical lenses like the Biogon may not have this issue.

Although to use it underwater Hasselblad commissioned Zeiss to design and make a Ivanoff corrector port and dioptre to make the best use of the Biogon 38mm lens underwater. Rather than a dome.

Although I have never tested it (I do own the Ivanoff, dioptre and Biogon 38mm on a SWC), my suspicions are in agreement with Paul that the Biogon would not measure up against most modern lenses. The instant review of digital has made us incredibly demanding about corner sharpness - in a way that we weren't 10 years ago.

Alex

Hi Tom/Alex

Tom, I'm not absolutely certain that I follow your comments but I'd suggest that my quoted discussion does assume an awful lot and is 'true' only with many assumptions made (ie its an overview of the simplistic theory). As subject matter becomes closer, a whole lot of discrepancies will certainly start to appear, most of which will add to image degredation. The whole point of my explanation is to show that whlst empirical testing will establish an optimised system, no amount of adjustment and testing can get around the deficiencies of a simple concentric dome port in front of a lens approaching 90 degree field of view. At some point a sensor and lens combination will have its corner performance limited by any dome port and I'd say Troy is probably using a system which will be so limited, as will most hires dSLRs and ultra wides - this is now the holy grail of underwater photography and will probably require innovative design to deal with.

On the comment about the Biogon design, I have no doubt that it is good (I've owned a couple myself) but manufacturers are now having to redesign and optimise lenses for digital. There are some interesting explanations of the abilities of Leica M lenses on digital but even these most beautifully designed, hand assembled pieces of glass have problems with digital (one of mine is in Germany being recallibrated for digital at the moment!). None-retrofocus designs w/a designs may cause s many problems as they solve (the M sensor uses angled micro lenses to solve some of them and the camera uses some form of image processing too).

Alex, I met someone camping out in the bush recently, miles from anywhere - tent, compressor, generator and diving gear plus Hasselbld housing!

Although to use it underwater Hasselblad commissioned Zeiss to design and make a Ivanoff corrector port and dioptre to make the best use of the Biogon 38mm lens underwater. Rather than a dome.

 

Although I have never tested it (I do own the Ivanoff, dioptre and Biogon 38mm on a SWC), my suspicions are in agreement with Paul that the Biogon would not measure up against most modern lenses. The instant review of digital has made us incredibly demanding about corner sharpness - in a way that we weren't 10 years ago.

 

Alex

The Petersen-Gates H38 housing uses a dome port. Without a proper digital sensor for the superwides we will never know. 'Proper' might entail some of what Leica has employed in the M8. Rumors are that a FF M8 sensor may be on its way so there may be a new solution the using short BF lenses.

Tom

Hi Tom/Alex

 

Tom, I'm not absolutely certain that I follow your comments but I'd suggest that my quoted discussion does assume an awful lot and is 'true' only with many assumptions made (ie its an overview of the simplistic theory). As subject matter becomes closer, a whole lot of discrepancies will certainly start to appear, most of which will add to image degredation. The whole point of my explanation is to show that whlst empirical testing will establish an optimised system, no amount of adjustment and testing can get around the deficiencies of a simple concentric dome port in front of a lens approaching 90 degree field of view. At some point a sensor and lens combination will have its corner performance limited by any dome port and I'd say Troy is probably using a system which will be so limited, as will most hires dSLRs and ultra wides - this is now the holy grail of underwater photography and will probably require innovative design to deal with.

 

On the comment about the Biogon design, I have no doubt that it is good (I've owned a couple myself) but manufacturers are now having to redesign and optimise lenses for digital. There are some interesting explanations of the abilities of Leica M lenses on digital but even these most beautifully designed, hand assembled pieces of glass have problems with digital (one of mine is in Germany being recallibrated for digital at the moment!). None-retrofocus designs w/a designs may cause s many problems as they solve (the M sensor uses angled micro lenses to solve some of them and the camera uses some form of image processing too).

 

Alex, I met someone camping out in the bush recently, miles from anywhere - tent, compressor, generator and diving gear plus Hasselbld housing!

Think water-contact optics (re. Nikonos RS).

What sort of calibration? Even with film the real fast Leica lenses (such as Noctilux f/1.0) had to be calibrated to specific bodies because the tolerances are so low.

Tom

All ports are "water contact optics". Video systems use solutions more elaborate that dome and flat ports as a matter or routine. They have to although their resolution requirements, until recently, aren't nearly what still cameras have.

What is meant by the jargon 'water contact optics' or more specifically a 'water contact lens' is a lens designed from the ground up to be used in water. It is not suitable to use such a lens in air. Adapting an air lens to be used UW is not the same thing. The Nikonos lenses with the W and RW designation are water contact lenses. The 50mm lens for the RS works (one can form an image in the film plane) out of water (I have done this) but is not as sharp as UW, the corners are soft.

Tom