Underwater Optics Myths?

[Interceptor121]

I am pasting the links otherwise they will convert in low resolution

 

https://interceptor121.files.wordpress.com/2024/05/20240506_mf200078_canon-15mm-f11.jpg

 

https://interceptor121.files.wordpress.com/2024/05/20240506_mf200109_wwl1-f11.jpg

 

https://interceptor121.files.wordpress.com/2024/05/20240506_mf200094_sony-1635-f11.jpg

 

I would like to understand where does the myth that a fisheye lens has good edges come from? Fisheye lenses have horrible edges and yes they do improve a bit with a dome but not that much

Water contact optics look good however at close range they look more 28mm than 130 degrees can't really be used for CFWA as they loose field of view

 

Rectilinear lenses properly set up beat the lot if the lens is good

 

Are we getting into some sort of propaganda post truth reality?

 

[Adventurer]

4 hours ago, Interceptor121 said:

https://interceptor121.files.wordpress.com/2024/05/20240506_mf200094_sony-1635-f11.jpg

Can you verify this is @ 16mm zoom position?

 

Did you host the fisheye in a full hemisphere dome or rape it in your 230mm zen dome? Maybe you can elaborate a little more and fill us in on the details. 
 

I must say that in your blog you were previously very positive about the Canon 15mm and Canon 15mm Kenko 1.4x TC. So I am surprised to see your latest sink shots look like this with it.

 

Generally there is a high chance of having to live with a mispositioned Canon 8-15mm Fisheye. Without full sphere you can almost be sure to throw away some light rays. The “Fisheye Bonus” cannot buy back the misalignment of the entrance pupil. Therefore I will try the old SIGMA 15mm EZ DG f2.8 Fisheye and not the Canon. 

 

I hope this does not hijack the thread but I’d love to read more about your previous teasered upcoming blogpost:

 

” For the purpose of this article I will consider only underwater imaging, split shots and over and under have different considerations and will be addressed separately in due course. “
 

source: https://wp.me/p2QoIB-k3J

 

Maybe we can touch that quote in the course of this thread and also work out the better split shot lens?
 

Rectilinear vs Fisheye

[Interceptor121]

None of those shots have teleconverter

one is fisheye 15mm the other is wwl-1 and finally 16-35mm

i used the zen for the fisheye and 16-35 to give an idea of field of view 

 

the edges of a fisheye look bad and actually the tc doesn’t deteriorate them 

fisheye have poor edges nothing is going to change that

[Adventurer]

51 minutes ago, Interceptor121 said:

fisheye have poor edges nothing is going to change that

By using the 230mm dome you have amplified the optical error and I am really surprised you did as from the knowledge published and rehearsed on your website you certainly know better.

 

I have some of my sharpest shots, including sharp corners with a fisheye behind a correctly positioned full hemisphere dome.

[Interceptor121]

24 minutes ago, Adventurer said:

By using the 230mm dome you have amplified the optical error and I am really surprised you did as from the knowledge published and rehearsed on your website you certainly know better.

 

I have some of my sharpest shots, including sharp corners with a fisheye behind a correctly positioned full hemisphere dome.

It does not make any difference in the centre the only impact is chromatic aberration at the extreme edges

 

Fisheye lenses have blurry edges that is the reality. A dome mildly improves edges especially if you are not close but at the end they are terrible and nothing will change that. Mostly you don't notice it because the edges are dark but the edges are crap 

 

Stopping down improves matters but you need to hit really high values like f/14 and similar to see some effects and by then centre performance has gone

Edited May 7 by Interceptor121

[Interceptor121]

This is an image with the 140mm dome at similar fov the differences are minimal and mostly in the centre the edge are actually a tiny bit worse but there is less CA

https://interceptor121.files.wordpress.com/2024/05/20240506_mf200082_canon-15mm-f11-140mm.jpg

 

[Interceptor121]

With regards to the water contact optics those are different comparision with dome or flat ports are not possible as those are demagnifiers.

Topic to be covered separately on why they work better in some cases

[Adventurer]

Mhh,.. strangely this does not copy my experience with fisheyes in the past.

 

I have to look more deeply into this soonish and re-verify.

 

I can just imagine right now that your sink test shot is extremely stressing the CFWA end and close focus of your Fisheye and therefore looks so disappointing.

 

when quickly comparing optical bench hub I see find also that you cannot really perfectly position the 8-15 canon plus adapter on the npp without vignette? I think you are off by a few cm.. which ist substantial. Am I right quickly browsing this and running the numbers?

[Interceptor121]

1 hour ago, Adventurer said:

Mhh,.. strangely this does not copy my experience with fisheyes in the past.

 

I have to look more deeply into this soonish and re-verify.

 

I can just imagine right now that your sink test shot is extremely stressing the CFWA end and close focus of your Fisheye and therefore looks so disappointing.

 

when quickly comparing optical bench hub I see find also that you cannot really perfectly position the 8-15 canon plus adapter on the npp without vignette? I think you are off by a few cm.. which ist substantial. Am I right quickly browsing this and running the numbers?

 

 

The tiny irrelevant vignette on the right side is due to the fact that the lens might have moved which in turn would help the situation if it was 14.9 instead of 15mm nothing to do with position you will discover housing and ports are not aligned with lenses but this is another story

Either way that is with the zen using 35mm instead of 30 in an attempt to position it properly the lens is way off where it should be 

the last shot is 140mm and yet no improvements

The sink test stresses CFWM however as infinity is at 30-50cm shooting far will not be a life change and moreover will benefit any lens

Fisheye are so bad that the dome compression helps the poor corners but they are not good whatever way you look at them in a cfwa image. I have looked in all my images carefully the classic example is the fish that is on a fan at first sight all is good then you look closely and the one in the edge is blurred and not because of depth of field just because the fisheye lens is weak at the edges. This is something any photographers knows but in the underwater world is like the opposite happens because there is a dome but that is not specific to that lens

 

 

Edited May 7 by Interceptor121

[ChrisH]

It might be a little Off topic, as it is not directly about the corner performance, but the fisheye lens is mostly used because of its wide angle of view and ability to focus very close. So you can get very close to your subject and shoot through the least amount of water possible. That is where image quality underwater can be gained the most: shooting through as little water as possible. With the normal (rectilinear) wide angle lens you need to shoot from a much more distance to get the same subject in the frame. if you shoot all lenses from the same distance, it will not show the real  benefit of the fisheye lens. In reality you would have to back off a lot with the rectilinear lens to get the same subject in the frame as with the fisheye, giving away image quality due to shooting through more water.

 

I just wanted to state this, as the initial post could lead to the impression that fisheye lenses provide poor image quality underwater, which is not correct in my opinion.

[Interceptor121]

11 minutes ago, ChrisH said:

It might be a little Off topic, as it is not directly about the corner performance, but the fisheye lens is mostly used because of its wide angle of view and ability to focus very close. So you can get very close to your subject and shoot through the least amount of water possible. That is where image quality underwater can be gained the most: shooting through as little water as possible. With the normal (rectilinear) wide angle lens you need to shoot from a much more distance to get the same subject in the frame. if you shoot all lenses from the same distance, it will not show the real  benefit of the fisheye lens. In reality you would have to back off a lot with the rectilinear lens to get the same subject in the frame as with the fisheye, giving away image quality due to shooting through more water.

 

I just wanted to state this, as the initial post could lead to the impression that fisheye lenses provide poor image quality underwater, which is not correct in my opinion.

 

Fisheye lenses have poor edge resolution that is a fact

Of course they also have very wide field of view which is useful if you need to get close

Some old fisheye lenses did not focus close requiring large domes. Some more recent ones focus at 15cm from the sensor plane

On the same note most rectilinear wide angle lenses focussed at 30cm and more but mirrorless focus many times at 25cm and some even less 19 or 22 cm

This means the gap is getting much narrower.

Obviously fisheye lenses have a wider field of view that helps composition however the area where the resolving power is very high is limited to the APSC crop in a full frame rig. There is also another factor for close focus shooting rectilinear lenses have perspective distortion which is not pleasant so although they may have actually more resolution the resulting image may not be something you like

This does not mean that fisheye lenses have more depth of field for example as depth of field depends on focal lenght and magnification nor that fisheye lenses are sharp in the corners

There is an obvious use case for fisheye lenses because they make the centre of the frame pop however on the same account this is not very useful if you have something interesting off centre and neither means the edges look good when they do not

When you compare the APSC crop of a fisheye lens and a 16mm lens they are the same however with a fisheye you have the space for a (blurry) background while with rectilinear you would fill the frame

 

I am not playing advocate for rectilinear lenses vs fisheye I have both but I want to highlight that certain perceived benefits indeed do not exist 

In some cases visibility is poor you need to get close and fisheye lenses help a lot otherwise we would all be using tele lenses underwater

 

Edited May 7 by Interceptor121

[ChrisH]

I personally could not recall a situation where I would not need to get very close to the subject underwater. And being able to do that is key. Thus the almost exclusively use of the fisheye lens for wide-angle pictures underwater. For video this is a different story of course. But I can count the (good) pictures I did with rectilinear lenses almost on two or three hands and often people won't even bring other lenses than fisheye for wide angle trips.

 

The only use cases for rectilinear lenses underwater would be in my opinion pool shots of models, inside caves (that don't suite the fisheye distortion) and sometimes wrecks. I personally don't mind using a fisheye for wrecks, but some people don't like the look (although often you can easy get shots that most people would not relate to a fisheye, because the fisheye distortion is not really pronounced). I would not ever use a rectilinear lens on a reef scene or something like that. If you need to shoot sharks that don't come very close you can use a rectilinear lens, but the quality will be poor just because you shoot through a lot of water.

 

I don't see poor corner performance in my fisheye shots, at least the shots would be much worse with rectilinear lenses. Of course that might be because I am not pixel peeping and mostly look at the overall impact of an image, so more scientific tests might be different. But then again, if it can't translate to "the field", the result of the scientific test might be of no use. That is of course only for me.

[Interceptor121]

30 minutes ago, ChrisH said:

I personally could not recall a situation where I would not need to get very close to the subject underwater. And being able to do that is key. Thus the almost exclusively use of the fisheye lens for wide-angle pictures underwater. For video this is a different story of course. But I can count the (good) pictures I did with rectilinear lenses almost on two or three hands and often people won't even bring other lenses than fisheye for wide angle trips.

 

The only use cases for rectilinear lenses underwater would be in my opinion pool shots of models, inside caves (that don't suite the fisheye distortion) and sometimes wrecks. I personally don't mind using a fisheye for wrecks, but some people don't like the look (although often you can easy get shots that most people would not relate to a fisheye, because the fisheye distortion is not really pronounced). I would not ever use a rectilinear lens on a reef scene or something like that. If you need to shoot sharks that don't come very close you can use a rectilinear lens, but the quality will be poor just because you shoot through a lot of water.

 

I don't see poor corner performance in my fisheye shots, at least the shots would be much worse with rectilinear lenses. Of course that might be because I am not pixel peeping and mostly look at the overall impact of an image, so more scientific tests might be different. But then again, if it can't translate to "the field", the result of the scientific test might be of no use. That is of course only for me.

Or maybe your rectilinear set up is poorly put together which is true for majority of people out there complaining about blurry edges?

[ChrisH]

18 minutes ago, Interceptor121 said:

Or maybe your rectilinear set up is poorly put together which is true for majority of people out there complaining about blurry edges?

Yes of course, I cant exclude that. I also only used the Nikon AF-S 16-35 F4, which had the reputation of beeing the least worst option for Nikon rectilinear underwater. I didn‘t test any of the new alternatives for Z-Mount. 

 

But again, even with better optics, the result would be the same: for the same shot, i would have needed to back off a good amount compared to the fisheye! And the overall image quality would thus suffer compared to the fisheye. 

 

Also my fisheye shots don‘t look anything like the test shots you did. The quality in the corner of the frame is vastly better.

[CaolIla]

Since the last hollidays I buy the new Canon RF 10-20 mm I only shot in lac for the moment I'm happy I can go "very close" of the subject... in a short time I'll try it on the reef of Bunaken ... 

 

I'm a lover of rectilinear lense..  but when Canon will propose a RF fisheyes i'll probably try it... and after that I'll be able to compare .

 

 

[John E]

Maybe it is worth a test in a pool and have the same composition i.e backed off on the non-fisheye - this would better reflect real world use - as Chris H points out about the shooting distance for a given subject. For the Canon 8-15mm fisheye, if you are talking about an APS-C crop wouldn't it also be relevant to try it wider than 15mm? (noting the Sony A7Rv has the same pixel density as the A6700). i.e. I wonder what is the fisheye pictures edge difference between the APS-C crop at around 10mm (or whatever is the exact equivalent) to full frame at 15mm?

 

Then there is the question of the view at the fisheye edge being so oblique anyway. 

 

 

[Interceptor121]

4 hours ago, John E said:

Maybe it is worth a test in a pool and have the same composition i.e backed off on the non-fisheye - this would better reflect real world use - as Chris H points out about the shooting distance for a given subject. For the Canon 8-15mm fisheye, if you are talking about an APS-C crop wouldn't it also be relevant to try it wider than 15mm? (noting the Sony A7Rv has the same pixel density as the A6700). i.e. I wonder what is the fisheye pictures edge difference between the APS-C crop at around 10mm (or whatever is the exact equivalent) to full frame at 15mm?

 

Then there is the question of the view at the fisheye edge being so oblique anyway. 

 

 

I do plenth of tests in the pool as per my website and I am not using APSC crop

What I am saying is that the area that has resolution in the lens broadly corresponds to the APSC crop, anything outside is garbled by the distortion

If you use a 10mm lens with APSC sensor you have a gain od=f depth of field but this has not to be confused with resolution measured on a flat target which is what I and also Nauticam do

Real world can't be tested it is not reproducible and there are other factors in play which are a separate conversation however when it comes to dome ports impact all optics equally fisheye or not

[Klaus]

I had never heard of the myth that fish-eyes have good corner performance. Top-side tests usually conclude that the field of view is too large for the test charts available (I.e. the patterns don’t extend into the very corners because of distortion and AOV), „real world“ shots have a chromatic aberrations which of course can be better or worse depending on the lens. Most of the times, the tests conclude with something like „ … fisheyes are extreme lenses with inherent compromises.“

Now underwater, it‘s a different story. The distance argument stated above is what I have read pretty much exclusively as the argument for using a fish-eye and it pairs well with the fact that the distortion is not as obvious for most UW subjects. 
There is of course nothing wrong with discussing very special aspects of optical setups - but I had never read about the supposed myth that started this thread. 

[Adventurer]

22 minutes ago, Klaus said:

but I had never read about the supposed myth that started this thread.

Me neither, but it’s easy to agree to as there is something else often pointed out in their relation with dome ports.

 

“Their circular crafting makes them more compatible with the curved shape of the dome and projected virtual image, when compared to rectilinear lenses.”

[Interceptor121]

16 minutes ago, Adventurer said:

Me neither, but it’s easy to agree to as there is something else often pointed out in their relation with dome ports.

 

“Their circular crafting makes them more compatible with the curved shape of the dome and projected virtual image, when compared to rectilinear lenses.”

There is a very mild improvement of the same fisheye lens in water compared to air as result of the dome bringing more depth of field fisheye lenses have an obvious issue of curvature going in the opposite direction

However on the same coin depth of field also helps a rectilinear lens as this is the way you cure field of curvature

Ultimately those shots are at very close range where those effect should be apparent yet at f/11 the rectilinear super wide lens has actually better edges which means the above statement may be true but does not make the fisheye lens better in that regards it is indeed worse

If you read some articles on line you also read other fantasies like fisheye have more depth of field? and other tales

 

With modern mirrorless lenses there is an aperture size that cures the spherical aberration (this indeed applies also to all lenses) from what I can see this is around f/5.6 to f/8 depending on port and lens. To say that you need to stop to f/16 to get good edges is definitely NOT true. You may need f/16 in certain situations for depth of field but if your port design is poor you can go all the way to f/22 and achieve nothing

Edited May 8 by Interceptor121

[Interceptor121]

7 hours ago, ChrisH said:

Yes of course, I cant exclude that. I also only used the Nikon AF-S 16-35 F4, which had the reputation of beeing the least worst option for Nikon rectilinear underwater. I didn‘t test any of the new alternatives for Z-Mount. 

 

But again, even with better optics, the result would be the same: for the same shot, i would have needed to back off a good amount compared to the fisheye! And the overall image quality would thus suffer compared to the fisheye. 

 

Also my fisheye shots don‘t look anything like the test shots you did. The quality in the corner of the frame is vastly better.

That lens is terrible even on land and with a working distance of 29cm not even worth buying a port for it

The only lens of F mount that was decent was the 14-24 however the working distance of 28cm penalised it

 

 

Underwater photography has been lagging and seems still to continue with DSLR type approach to optics where the world meanwhile has moved on. The nikkor 8-15 was on a different planet compared to most rectilinear and focuses very close 

 

I do not know what goes on with Z mount however in the E-mount we have lenses with working distances of 19,22,25 cm this makes quite a difference and those lenses are way sharper than the old optics on the f mount. We are also penalised as there are no native fisheye options for mirrorless yet but adapter Canon or Nikon are still very valid options

 

This does not mean there that rectilinear lenses are an alternative to fisheye however for those shots were fisheye was too large modern mirrorless rectilinear lenses offer a real quality alternative and some of them do not need massive domes as they focus close. Even a massive dome is lighter than a 'compact' water contact optic to make another comparison

 

[John E]

So what is the real world interpretation? If one has the choice between Sony 26-60mm + WWL-1, the Canon 8-15mm, Sony 14mm or Sony 16-35mm at it's widest, and is taking the same type of shot such as a reef scene, which would be the better option for overall sharpness? Assuming the correct extensions are used and the appropriate dome size, is the fisheye the worst option unless the specific style is needed, or does the close working distance reducing the amount of water in front of the camera offset any drawbacks and mean it is the best option?

[Interceptor121]

2 hours ago, John E said:

So what is the real world interpretation? If one has the choice between Sony 26-60mm + WWL-1, the Canon 8-15mm, Sony 14mm or Sony 16-35mm at it's widest, and is taking the same type of shot such as a reef scene, which would be the better option for overall sharpness? Assuming the correct extensions are used and the appropriate dome size, is the fisheye the worst option unless the specific style is needed, or does the close working distance reducing the amount of water in front of the camera offset any drawbacks and mean it is the best option?

The field of view are not comparable. Specifically the 28-60 WWL-1 compares with the Canon 8-15 + TC and with the 14mm

The 16-35 is a bit narrower and the fisheye is obviously wider

 

Personally I use the Canon 8-15 with TC with the zoom gear because in many reef scenes there is nothing interesting in the corners and with this set up I can shoot super wide and zoom if required. The APSC central area is indeed sharp and that is typically what matters. I have also seen that the TC takes nothing away from IQ and therefore I have it as default

I would not use any of the WWL or rectilinear lenses for reef scenes

 

I am not in a crusade against fisheye in fact it is the first lens anyone should get

[Klaus]

Image quality aside (and I know this is now off-topic), the shots I admire most ist where the fisheye distortion is used as an element of the composition - a feature, not a bug. I can mostly remember this from CFWA, not so much from reefscapes. But I guess these occasion are not that frequent, hence zoomable options are more flexible. Just a thought on aspects beyond the pixel details.

[Interceptor121]

7 minutes ago, Klaus said:

Image quality aside (and I know this is now off-topic), the shots I admire most ist where the fisheye distortion is used as an element of the composition - a feature, not a bug. I can mostly remember this from CFWA, not so much from reefscapes. But I guess these occasion are not that frequent, hence zoomable options are more flexible. Just a thought on aspects beyond the pixel details.

the field of view of the fisheye lens and the barrel distortion magnify the centre of the frame and create space around it this is a unique feature

If you instead shoot a flat piece of reef with a lot of stuff from centre to edges any lenses works the same it is even hard to tell how it was shot

This image for example could you tell the lens it was taken with?