I’d like to throw out a structured thought experiment and get input from others who are chasing maximum optical quality underwater. The idea is a bottleneck search: identifying which part of the optical chain ultimately caps achievable sharpness, and how modern lenses may have shifted that ceiling.

If we ignore water quality for a moment, I think underwater image quality can largely be reduced to three primary, controllable factors:

1. Camera & Sensor
   Pixel pitch and diffraction limits define the aperture range where the sensor can still resolve real detail. Higher-resolution sensors tend to hit diffraction earlier, which already caps usable sharpness.

2. The Lens (Topside Performance)
   Every lens has a measurable resolution ceiling, typically expressed in line pairs per millimeter (lp/mm) under controlled lab conditions. This is the absolute upper bound of what the system can ever deliver.

3. Dome Port or Underwater Corrective Optic (WACP, FCP, etc.)
   These elements can preserve or destroy resolution, but they cannot exceed the lens’s native resolving power. They are enablers—not multipliers.

My working assumption is therefore:

No underwater optic or dome can ever push image quality beyond what the lens–sensor combination can already resolve topside.

_{For anyone interested in comparing topside lens performance in a more objective way, I’ve found two resources particularly helpful. The ISO 12233 sample crop tool at}

_{https://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=1624}

_{allows direct visual comparison of resolving power across lenses and sensors, while}

_{https://opticallimits.com/the-list/}

_{provides a concise overview of laboratory-measured resolution data across many modern and legacy lenses.}

A Reference Point: Canon EF 8–15mm Fisheye

In underwater photography circles, the Canon EF 8–15mm f/4L Fisheye has long been regarded as one of the sharpest and most reliable underwater lenses ever made. That reputation is not anecdotal—it is backed by lab data.

From LensTip’s resolution test (more than a decade old):

“The performance you can see on the graph can be only described by one word: revelation. At shorter focal lengths the lens, even wide open, reaches a level of 50 lp/mm or higher and by f/5.6 it gets to record-breaking values of 53–54 lp/mm.”

_{( source https://www.lenstip.com/311.4-Lens_review-Canon_EF_8-15_mm_f_4_L_Fisheye_USM_Image_resolution.htm )}

At the time, this effectively outperformed almost every other EF lens tested. In underwater use, the lens also benefits from extremely close focusing, which reduces water column and often compensates for other optical weaknesses. This is arguably a fourth, indirect factor: minimizing the amount of water between lens and subject.

The Shift with Modern RF Lenses

Here is where things get interesting.

If we look at modern Canon RF lenses, even non-L designs now meet or exceed resolution levels that were once considered exceptional. For example, laboratory testing from digitalkamera.de reports the following for a very modest kit zoom:

Canon RF 24–50mm F4.5–6.3 IS STM (tested on a 24 MP EOS R8):

“The lens reaches a maximum resolution of 57 line pairs per millimeter (lp/mm) in the image center and 49 lp/mm at the image edge at a focal length of 35 mm.”

In other words:

A small, inexpensive, non-L RF zoom already exceeds the peak resolution of the legendary EF 8–15mm fisheye, at least in topside lab conditions.

This strongly suggests that modern mirrorless lens design has raised the baseline for optical sharpness, and that many assumptions carried over from the EF era may no longer hold.

The Exception—and the Real Question

The EF 8–15mm remains special not just because of resolution, but because it combines:

• Very high lp/mm performance

• Extreme close-focus capability

• Excellent compatibility with small domes

• Reduced water column in real underwater scenarios

That combination makes it a rare exception where practical underwater sharpness can rival or exceed newer lenses that are sharper on paper but harder to deploy underwater.

This leads me to the real questions I’d like to discuss:

• What is the sharpest lens (in lp/mm terms) available today for your system—Canon RF, Canon EF, Nikon Z, Sony E?

• Which of these “topside gold nuggets” can realistically be used underwater with domes or corrective optics like WACP/FCP?

• Are we still overvaluing classic underwater lenses, while underestimating what modern mirrorless optics could deliver if paired with the right underwater solutions?

I’m very curious to hear which lenses you think currently define the true optical ceiling for underwater photography in your system—and whether anyone has successfully translated that topside sharpness into real underwater results.

Edited Saturday at 11:53 AM2 days by Adventurer

How about which lenses underwater are excellent underwater but are horrible topside? Tokina 10-17, anyone?

21 minutes ago, TimG said:

How about which lenses underwater are excellent underwater but are horrible topside? Tokina 10-17, anyone?

Hi Tim,

I hope we do not drift away from my initial aim with this post. I think I pretty much highlighted the practicability part above and can tell you (by experience) that the Tokina 10-17 pitched a great concept but is IQ wise totally inferior to the Canon EF 8-15 F4 L - for the sake of trying to nail the cream of the crop here, I would also appreciate if we can stick to FULL FRAME MIRRORLESS systems and leave APS-C, M43 and compact out of the topic.

My bad

13mm RS on sony mirrorless is (imho) by far the sharpest lens underwater. the resolution it resolves is spectacular and it surpasses thr 8-15 from canon or nikon in all the categories mentioned

Hi,

I don`t want to go too deep. Just give me a definition of "sharp". Do I have a "sharp" photo, just with a so called sharp lens.

@ Adventurer: From my point of view I`m not clearly shure what you would like to compare. Only mirrorless lenses or also full frame DSLR lenses which can be used with adaptor. By the way, who can let me know the OPTICAL difference using full frame f.e. 8-15mm lens on a mirrorless or a Dslr system.

FYI: Fisheyelens I use 13mm RS & 8-15mm on Nikon DSLR & Im lucky with the results.

BrMarkus

2 hours ago, MatthewSullivan said:

13mm RS on sony mirrorless is (imho) by far the sharpest lens underwater.

Yes, indeed - glad you mention it, Matthew. However we all rely on personal impression reports like yours and Alex‘s. Unfortunately there has been no quantified lab test with this lens which measures (underwater) in lp/mm. So we do not know how much we are potentially missing out 😅. I do believe you guys, though!

To give everyone a good, relatable reference: the often-discussed Canon RF 100mm f/2.8L Macro manages to reach up to 86.5 lp/mm at f/4 in the image center. (source: digitalkamera.de lab tests). So unless you use focus stacking it‘s not very realistic to archive under practical underwater application. @F16 it bends down to 61,8 lp/mm in the center; and with F22 it’s down to 49,5 lp/mm. So due to macro practical depth of field requirements you might have experienced that lens just similar sharp or even less sharp than your EF8-15. I recall that @MatthewSullivan was praising the RF100 macro for its sharpness in the underwater photography show on YT and also happens to shoot and own the adapted RS13 fisheye ( „the holy grail“ ). So having the numbers rankings, he could maybe make a well informed expert guess on the RS13‘s lp/mm sharpness value?

1. RS13 adapted = ???? lp/mm

2. RF100 @ F4 = 86,5 lp/mm

3. EF 8-15 Fisheye = 54 lp/mm

4. RF 24-50 @ 24mm + F11 = center 52,8 lp/mm edges 41,2 lp/mm

@Muellema to answer your question: all lenses that can be adapted / used with full frame mirrorless; or better: that make sense of being used to max out IQ on your full frame mirrorless system. Sharp RAW output is measurable in lp/mm or lw/ph (resolution).

Edited Saturday at 05:11 PM1 day by Adventurer

Wetpixel.com

Test: Optical performance of Nikonos 15mm, flat and dome...

Wetpixel is the premiere community website dedicated to underwater photography and videography

12 hours ago, Adventurer said:

If we look at modern Canon RF lenses, even non-L designs now meet or exceed resolution levels that were once considered exceptional. For example, laboratory testing from digitalkamera.de reports the following for a very modest kit zoom:

Canon RF 24–50mm F4.5–6.3 IS STM (tested on a 24 MP EOS R8):

“The lens reaches a maximum resolution of 57 line pairs per millimeter (lp/mm) in the image center and 49 lp/mm at the image edge at a focal length of 35 mm.”

In other words:

A small, inexpensive, non-L RF zoom already exceeds the peak resolution of the legendary EF 8–15mm fisheye, at least in topside lab conditions.

This strongly suggests that modern mirrorless lens design has raised the baseline for optical sharpness, and that many assumptions carried over from the EF era may no longer hold.

Isn't this the same lens that @DreiFish tested and pronounced to have pretty poor optical quality in his thread from a little while back. Consulting google there are some tests that pronounce optical quality to be quite good and others such as the digital picture tests with sample images that pronounce it to be quite poor. I agree there are some very good lenses coming out now, but you still tend to get what you paid for. The searching of reviews suggests that there is a lot of sample variation in this lens so it's a bit of a lottery it would seem about how good the lens you get is. Or perhaps the good reviews received hand picked lenses to test??

BTW the test for the the 24-50 on digitalkamera.de seems to be behind a paywall.

While I agree that having a sharp lens is great and certainly an interesting exercise to trying to find the sharpest you can for your UW photography and do not in any want to discourage your quest, when under water IMO flexibility is quite important as you can't swap to a more optimal lens like you can on land. Never mind the Ansel Adams quote:

"There is nothing worse than a sharp image of a fuzzy concept"

17 hours ago, Grantmac said:

Wetpixel.com

Test: Optical performance of Nikonos 15mm, flat and dome...

Wetpixel is the premiere community website dedicated to underwater photography and videography

Thanks for this Link!

Surprisingly, this puts the “legendary” Nikonos V / UW-Nikkor 15mm at about 51.9 lp/mm, i.e. below the RF 100mm and closer to EF 8–15mm land performance — and here’s why:

1) The test is not full frame (crop factor first, and it also explains the reduced underwater FOV)

The Wetpixel/Achtel measurements were done on a RED EPIC MYSTERIUM-X (Super-35) recorded in 5K 2:1 (5120×2560). In that mode the active imaging area is roughly 27.65 × 13.83 mm, which is about a 1.40× diagonal crop compared to 36×24 mm full frame.

That crop factor directly explains why the underwater diagonal FOV quoted for the Nikonos 15mm looks “reduced”: the UW-Nikkor 15mm is specified around ~94° diagonal underwater, and applying ~1.40× crop gives roughly ~75°, consistent with the ~74° figure discussed in this context. So this is primarily a format/crop effect, not the lens “getting narrower” underwater.

2) Why an “edge” Imatest ROI on this setup is not a full-frame corner

If Imatest samples an edge/ROI near the edge of the RED frame, that location maps to about 1 / 1.40 ≈ 0.71 of the full-frame image radius. In other words: it corresponds to mid-to-outer field on full frame, not an extreme 36×24 corner. So the Imatest result should not be interpreted as “full-frame corner performance.”

3) The lp/mm number is a derived value (LW/PH → lp/mm conversion)

Imatest reports MTF50 in LW/PH (line widths per picture height). To convert to lp/mm on the sensor plane:

• lp/mm = (LW/PH) / (2 × picture height in mm)

For RED 5K 2:1, picture height is ~13.83 mm. With the reported MTF50 ≈ 1434 LW/PH:

• lp/mm = 1434 / (2 × 13.83) = 1434 / 27.66 ≈ 51.9 lp/mm

Bottom line: the widely quoted ~51.9 lp/mm is a converted sensor-plane MTF50 number on a cropped Super-35 capture, and the sampled field position is closer to mid/outer field than a true full-frame corner.

As the values I initially posted for comparison were also center / near center based performance values in lp/mm we can use this for comparison. We should however honor the high open aperture resolution performance of the Nikonos V 15mm while keeping the dismal flexibility and manual focussing required underwater in mind when judging the practicability.

Just want to add one thought. We shoot a whole system / configuration. A great lens can be sheit under water. Nikon 14-24 anyone? So while a lens won’t get sharper or better under water, it might not be as degraded as others.

Comparing land performance figures against an UW lens seems less than valid no?

I'd like to see how it performs on a FF sensor in a controlled test. I don't personally have the time, infrastructure or desire to do so but I am curious.

I have the lens but have yet to be able to take it underwater.

While this is certainly an interesting exercise, there is a fundamental problem. You will see this notice on sites like optical limits:

Please note that the MTF results are not directly comparable across the different systems!

what does this mean - it means they are testing the combination of the lens, sensor, anti alias filter, de-mosaicing algorithm and any camera specific changes. But what does this actually mean? Well searching around I found a website that tested the same lenses on different camera systems. They tested a SONY 50mm f1.2 on both a Sony and Nikon camera and on SONY (A7RIII) its peak Imatest score was 3611 at f2.8 while on Nikon (z7) it was 4128 @ f2. There is also a test of a 35mm f1.4 EF lens on SONY, Canon and Nikon. The Canon camera used has an anti alias filter so returns significantly lower scores on the same lens Here is the site:

https://photographylife.com/our-canon-and-sony-lens-reviews-will-have-imatest-data-comparable-to-nikon

So where does this leave us? We can certainly work out the fastest knife in the drawer in any particular brand of lens if they are tested on the same camera. We know that a 3600 score will pretty much certainly sharper than say a 2500 score, but unless the top contenders are tested on the same system we will struggle to decide which is first place. We could use the plots supplied in the link above to approximately convert between the systems, however they only really apply to the cameras they use for testing.

As for the Nikonos test the biggest problem I expect is the fact that the testing was done on a completely different system. The centre score in LP/mm should still be valid for a full frame sensor with the same characteristics - the corners of course are a different story. And to be truly comparable the Nikonos is tested UW, you would need to test the other lenses UW as well to compare them. Then there is the problem of different amounts of water between the lens and target. You could probaly validly compare the 8-15 with a Nikonos 13mm equivalent but perhaps not a 16-35 lens for example.

Unfortunately we are drifting away in this thread with a lot of speculation, hairsplitting and nothing new brought to the table by some contributors.

This was not what I wanted in the topic

opener. I want lens pitches ( „potentials“ ) which we can explore and test AND quantifiable (lp/mm and LW/PH) to get an idea where they possibly stand. These of course have to be ranked in separate lists for each camera brand bayonet and are not cross comparable. It‘s also problematic if you compare testing website source A with website B or have to compare underwater or topside values. But you have to start somewhere to get a vague idea on what’s possible.

Also I think it‘s important to relate to something you know. Many of us have the EF 8-15 (or an RF100) and can therefore relate to how pixelpeeping a 54 lp/mm (or 86,5 lp/mm) candidate looks like.

Edited 6 hours ago6 hr by Adventurer

If only lens resolution and sharpness were the lowest hanging fruit for improvement of underwater images! Realistically, there are about a dozen things that are likely to have a greater impact on creating better images than lenses or camera sensors.

Sorry if that's not a helpful comment!

There is an aussie guy called Admiral Achtel who has done lots of resolution tests as he'd a specialist in high resolution cinematography (UW and on land)

https://achtel.com/underwater-cinematography/

He is someone who has spent lots of time measuring stuff and probably has the best information you are looking for (although he is also selling stuff - so be aware that the data he shares are likely to support what he sells).

As he (occasionally!) mentions his systems were used for filming Avatar. Which is a perfect example of what lots of people here are saying - you may have the sharpest lens - but you need more than that to avoid creating something dull to look at.

1 hour ago, Adventurer said:

Unfortunately we are drifting away in this thread with a lot of speculation, hairsplitting and nothing new brought to the table by some contributors.

This was not what I wanted in the topic

opener. I want lens pitches ( „potentials“ ) which we can explore and test AND quantifiable (lp/mm and LW/PH) to get an idea where they possibly stand. These of course have to be ranked in separate lists for each camera brand bayonet and are not cross comparable. It‘s also problematic if you compare testing website source A with website B or have to compare underwater or topside values. But you have to start somewhere to get a vague idea on what’s possible.

Also I think it‘s important to relate to something you know. Many of us have the EF 8-15 (or an RF100) and can therefore relate to how pixelpeeping a 54 lp/mm (or 86,5 lp/mm) candidate looks like.

Unfortunately that's internet forums for you people run off on tangents on things that interest them. But seriously - please continue to dig and find resources and test images to make your points.

14 hours ago, Chris Ross said:

what does this mean - it means they are testing the combination of the lens, sensor, anti alias filter, de-mosaicing algorithm and any camera specific changes. But what does this actually mean?

Well it turns out, that this is very unlikely to be a problem when searching the sharpest knives (lenses) for underwater imaging. As @Chris Ross 's linked blog post and @Alex_Mustard 's Admiral Achtel's thoughs brought up the optical Nyquist limit I dove a little deeper into the physics and technical aspects behind it. For Chris's concern,... if we look at MTF50 charts only, not MTF10/MTF20, the results are indeed quite comparable with a potential offset between 0% and -15%. We can keep this in the corner of our head.

IMPORTANT: the Nyquist limit is raging against Diffraction Airy disc limit.
Diffraction kicks in earlier then Nyquist limit, forcing you to use more conservative F-Stops.

But it is very handy to know what might define the true optical theoretical ceiling with various full frame cameras in 2026. For example my Canon R6 II will most likely not resolve lenses that excel beyond 83.33 lp/mm - and your Canon R5 II or Sony A1 II will most likely not suck out much more than that, as you will kill IQ earlier with diffraction.

I compiled the following handy table for your reference:

Edited 12 minutes ago12 min by Adventurer