The purpose of a Dome port?!?

Since I started with cameras I though dome ports purpose was to restore the air field of view of the camera and that do so the centre of curvature of the dome had to be placed near the entrance pupil of the lens.

Along the years I have started taking test shots of a ruler with an empty and full sink and I noticed in some cases considerable loss of field of view. I thought this was a trade off in those cases where the dome has to be kept to a reasonable size and you needed to avoid vignetting.

Typically entrance pupil information for lenses is not provided by manufacturers and furthermore for zoom lenses this moves on the lens itself. Recently I have acquired a full frame lens where this information was available and I have noticed that the suggested extension by the manufacturer was not on the entrance pupil but further behind. This made no sense to me as according to theory once you determine the entrance pupil the real point if you consider a thick dome is actually with the lens closer not further away.

I got curious about this situation so I started measuring the suggested position of the dome for a rectilinear wide angle lens I own. I then compared with the suggested manufacturer extension to find out the centre of the dome would be almost at the end of the lens. I then performed some tests to see if this was the entrance pupil and it was clear it was not and that the correct place was around 2 cm behind this point. I then enquired the manufacturer to understand why such long extensions are recommended and I found out that the design  for a dome port is pretty much like this

1. Select a given dome port large enough to give decent results

2. Put the camera and lens at the wide and and the dome on a macro pilot with the dome inserted in water pointing to a resolution chart

3. Perform various MTF measuremens moving the camera to determine the needed extension that does not vignette

In all cases I have checked this method gives a longer extension that the theory suggests this also means that the dome will actually loose field of view by definition.

I am not sure I agree with this approach for a number of reasons

1. If the field of view is being reduced to achieve sharpness I would like to know what I am loosing

2. Placing the lens further away from the dome results in the dome acting like a lens so it is clear it will be sharper but this could be achieved anyway zooming in that actually moves the entrance pupil backwards in most cases

3. The tests are only conducted at wide end this may be fine for a prime lens but for a zoom lens there is no guarantee that MTF will hold as you zoom in

Maybe I am being naive but if I buy a lens that has 100 degrees field of view and a dome I am expecting this to be more or less the same not to be cut down of an unknown amount based on how sharp are the corners of the image.

Am I alone? How many people know that domes may actually NOT restore the air field of view and not of a small amount?

I've come across two interesting POVs. One, is the camera lens is focusing on a "virtual" image produced by the dome wide angle lens. Secondly, I know an individual that builds Dome Wide Angle Lenses for most any housing. He calculates the "nodule" point for the lens in the housing and constructs the dome accordingly. I've used his domes successfully. However, I cannot afford his custom domes. They are top drawer and cost as much.

I shoot Zen wide angle lenses and have no problems with vignetting or sharp focus.

My understanding is that the dome to no restore field of view, but try to minimize the distortion generated by the water/air interface.

This distorsion increases with the angle of view, which is why a dome port is required with wide angle lenses. 

But it can only try minimize, as to completely optimize the distorsion, you would have to get a custom dome designed for each lens, and for a zoom lens you would only optimize a single focal length. This optimization requires careful positioning of the lens vs. the focal node of the port, and it restores some of the angle of view lost with a flat port. But I don't think that the purpose of the port is to restore fully the angle of view. In most case it would be impossible.

I expect the method used is more about expediency and finding the closest extension from the library of what is currently available in extension rings that will satisfy the majority of people who use the dome.  Most people would have a hard time determining if the lens is placed correctly according to theory but would immediately notice vignetting. 

I checked my Pany 7-14 which I use in a Zen Type II dome - this is a partial dome (not 180° curve) with a radius of 110mm.  that places the centre of curvature about 30mm behind the measured nodal point of the lens which is up near the front of the lens.  A bit of geometry shows that extending the dome much further would cause vignetting.  So in this case it a compromise to allow you to use a more compact dome with a larger radius of curvature than would be possible with a hemisphere dome.

@Algwyn distortion and field of view are related
I found this great model
https://oceanity.com.au/tools/dome-port-positioning/
In the case I discussed the entrance pupil is further away from the centre so you have pincushion distortion that reduces the field of view once corrected. So your 114 degree lens may become less than 110 practically
@ChrisRoss if what you say about the entrance of the 7-14 is correct the lens will be closer than ideal this would generate barrel distortion that actually increases field of view when corrected
Do you see this in your uncorrected raw images? The lens already has 5.2% barrel distortion so this would be apparent
However looking at the construction of the 7-14 I am not sure at all the entrance pupil is towards the front as looking at the lens schematic you can see that there are a number of converging lenses than a flat one located behind mid point and then diverging. Have you measured the entrance pupil with a pano head on a tripod or by hand?

Sent from my SM-A505FN using Tapatalk

did it with a tripod, levelled the clamp and head then moved it back and forth with a dovetail plate till the parallex shifts stopped, rotating it using the pan control.  I have seen other references that say the point is where the aperture appears to be when looking from the front and it does appear to be in the same vicinity.  I  estimated it was between the top of the MF ring and the line on the lens hood.  I may have completely stuffed it up but I recall I found a point where the parallex shift stopped, it was some time ago though.

8 hours ago, ChrisRoss said:

did it with a tripod, levelled the clamp and head then moved it back and forth with a dovetail plate till the parallex shifts stopped, rotating it using the pan control.  I have seen other references that say the point is where the aperture appears to be when looking from the front and it does appear to be in the same vicinity.  I  estimated it was between the top of the MF ring and the line on the lens hood.  I may have completely stuffed it up but I recall I found a point where the parallex shift stopped, it was some time ago though.

OK so if it was where you say it was it would be around 5 cm from the base considering the distance from the hood. The whole lens is 83mm and 25mm are in the housing so that should place the centre of curvature around 2.5 cm from the port which means the Zen dome should be 13.5 long excluding the part in the housing to be correct. I think the port is actually that long if not longer?

The dome is the N85 II and it is about 115 mm from the outer surface of the dome to the outer surface of the housing.  5cm is the correct measurement for lens flange to where I estimate for the nodal point.

@ChrisRoss so if the measurement of the entrance pupil was correct this port is 2 cm shorter and there would be barrel distortion and the field of view would be wider than air. Also need to consider that the camera body and lens are likely trying to correct this distortion already as the lens has some of it's own.
Also this is a thick dome so usually the real distance is corrected to a shorter one. Interestingly nauticam port chart for n120 has a 47 mm extension for the 180mm dome that had the dome centre 27mm behind so theirs would also be 5mm shorter and am pretty sure with the n120 neck it would not vignette easily
Do you see barrel distortion in your uncorrected raw files? Or a wider field of view?

Sent from my SM-A505FN using Tapatalk

I'll have to look into it, the lens does not get a great deal of use - so shots where I could pick that up are few.

I'll have to look into it, the lens does not get a great deal of use - so shots where I could pick that up are few.

So yesterday I put two lenses on a nodal slide and evaluated the required port length

Consistent to Nauticam port chart the nodal point at wide end was 1cm apart between the two lenses however in both cases Nauticam suggested extension is 1cm longer than would it be required 

Talking to their support they say this gives the best resolution even if it may not be the widest field of view

I have then looked at the characteristics of the lens, this is the leica 8-18, and there is corrected barrel distortion at wide end around 6% so what I think nauticam does is to pull the lens away the lens stops correcting the barrel distortion because the pincushion distortion induced by the longer extension compensates for it resulting in a longer extension.

This works well at wide end however as you zoom in usually the distortion goes away so it is likely that the performance of Nauticam choice will drop and genuine pincushion distortion will come into play they do not test this so it is not possible to say

It is also true that in a true dome the lens has to be a little bit closer than required anyway

So in short Nauticam recommended port for this lens is longer than expected to optimise sharpness at wide end at expenses of loss of field of view. They do not test the entire zoom range and the entrance pupil will move typically backwards making performance even worse

I cannot generalise this approach as there are lenses that have a physical construction where extending the port is not possible because of vignette (Panasonic 7-14mm) or a fixed hood (Olympus 7-14) so it is unclear what performance you are getting there but it is not possible to do otherwise

For the Panasonic 8-18 instead it is possible to perform an optimal position of the dome and if you wanted to do it according to conventional criteria you would use an extension 0.5 to 1 cm shorter