Underwater Camera Systems: A Comparative (in the Field) Performance Review

This was a weekend project, and it surely has some flaws and biases. Just giving our perspective on what we saw with these results.

Objective: Evaluate and compare three underwater video systems

RED Helium (7K RAW, 60 fps, Log; ISO 800; Canon 16–35 mm)
Canon R5 Mark II (4K sRAW, 60 fps, Log 2; ISO 800; Canon 10–20 mm)
Canon 1D X Mark II (4K Fine, 60 fps; ISO 800; Canon 24-70 mm)

all housed in Nauticam, with post production in DaVinci Resolve 20.

Locations & Conditions

• La Reina (Cerralvo Island, Mexico): Depth 2-5 m; visibility 25 m; open ocean blue; strong sunlight; friendly giant manta.

• El Vencedor (Cabo Pulmo, Mexico): Depth 14 m; visibility 10 m; turbid, greenish blue; high activity (8+ bull sharks).

Methodology

1. Standard Settings: All systems at ISO 800, auto (or custom) white balance, Log profiles (except Canon 1D X without Log). Focal length varied per dive.

2. Shot Pairing: Captured simultaneous or near simultaneous clips from two systems per scene; matched framing, distance, angle, and light as closely as possible.

3. Post Processing: DaVinci Resolve 20; primary correction (balance, contrast), secondary grading (hue/saturation), spatial denoise as needed.

Test Sequences

1. Giant Manta (Surface): Canon R5 II, RED: https://vimeo.com/1102366224

2. Manta + Diver: RED, Canon R5 II: https://vimeo.com/1102366387

3. Bullsharks: RED, Canon R5 II, Canon 1DX MII: https://vimeo.com/1102366448

Metrics & Rankings (for us)

1. Dynamic Range

1st: RED Helium — Exceptional latitude in harsh shadows/highlights; range ≈16 stops.

2nd: Canon R5 II — Robust performance in direct sunlight; range ≈14 stops.

3rd: Canon 1D X II — Limited by Rec.709 gamut; range ≈11 stops.

2. Sharpness (300 % zoom)

1st: RED Helium — Highest resolving power; artifacts negligible at 4K output.

2nd: Canon R5 II — Slightly softer than RED; no perceptible difference at normal viewing distances at 4K output.

3rd: Canon 1D X II — Lowest micro detail; adequate for 4K playback.

3.Color Science

1st: Canon R5 II — Accurate warm tone recovery with custom WB; maximum color grading latitude

2nd: Canon 1D X II — Natural skin tone palette; log less workflow simplified but limited latitude.

3rd: RED Helium — Neutral baseline; struggled with warm hues in turbid water; requires masking or advanced secondary curves especially in deeper dives.

4 .Noise

1st: RED Helium — Cleanest under low light; minimal spatial denoise applied.

2nd: Canon 1D X II — Moderate noise; less than R5 II in underexposed areas.

3rd: Canon R5 II — Most noise at base ISO 800; resolved via spatial denoise, recommended exposure +1 stop.

5. Ergonomics & Cost

1st: Canon 1D X II — Mid-weight/affordable (used market ~$5-7KUSD); ideal entry level.

2nd: Canon R5 II — Most compact; moderate cost ($~12KUSD); superior stabilization.

3rd: RED Helium — Highest cost (~$45K USD) and bulk; professional grade stabilization.

Conclusions

•RED Helium: Choose for maximal dynamic range and detail (e.g., commercial cinematography). Budget and weight are significant factors. Color grading will have to be taken care by a profesional.

•Canon R5 II: Best balance of image quality, color fidelity, and handling. Recommended for independent filmmakers and enthusiasts.

•Canon 1D X II: Cost effective entry to 4K underwater; ideal for novices or secondary camera.

After all those dives and countless shots from this weekend, it really comes down to being in the right place at the right time. No matter which camera you choose, if you’re closer, steady, and nail your angle and exposure, you’ll walk away with an amazing image. It’s recommended to prioritize positioning, timing, and stability—these elements often contribute more to a standout underwater image than the choice of camera, especially when viewing on a 4K display.

Technical Note:

Camera and codec choice is usually dictated by the SVOD or broadcaster, based on how well the camera’s image pipeline and recording format hold up under their platform’s compression. It’s not just about the look of the footage on set—it’s about how it survives multiple layers of encoding and streaming at lower bitrates. Platforms like Netflix, Apple TV+, and Disney+ often require primary cameras to capture at least 4K RAW or 10-bit 4:2:2 (RED Helium and R5 Mark II both have these codecs) with wide dynamic range and robust color depth, and favour intra-frame codecs (ProRes, XOCN, REDCODE) over Long GOP or 8‑bit 4:2:0. Ultimately, technical approval depends less on aesthetic appeal and more on how reliably an image holds up across global delivery systems.

Fantastic write-up and a huge thank you for putting in the time and effort to create this comparison!

Color is very subjective, but I've always thought that, out of the box, Canon's colors and their white balance (WB) underwater are subjectively more beautiful than other major brands, almost perfect as a starting point.

Initially, Sony's colors were the worst, but in recent releases, Sony finally has excellent color profiles for underwater shooting too.

A few technical points I'd love to discuss:

• For your custom white balance, were you setting it at depth using a slate? Getting accurate warm tones back can be tricky, and your results with the R5 II look great.
  

• Noise on the R5 II: Your finding about the R5 Mark II being the noisiest at base ISO 800 is interesting. If I understand correctly you were using SRaw/CLog 2?
  

• RED Color Science: Your comments on the RED's color science underwater match what I've heard from others. It provides a very clean, neutral starting point but requires a more experienced colorist to bring out natural water tones and skin tones compared to Canon's out-of-the-box science. Did you use RED's IPP2 pipeline in Resolve, and which LUTs, if any, did you use as a starting point?
  

• The 1DX II: It's fascinating to see the old 1DX Mark II still holding its own, especially on noise. That Motion JPEG codec is ancient but very robust. Its performance here is a testament to the idea that a high-bitrate, intra-frame codec (even if it's 8-bit) can sometimes be easier to work with than a more compressed 10-bit format if not exposed perfectly.

It would be interesting to see a more in depth head-to-head comparison between the RED Gemini and the Canon R5 Mark II, both shooting in RAW. 😇

Thank you again for you test

Sorry but we need a vimeo account to watch the video... from France I don't know for other country

36 minutes ago, CaolIla said:

Sorry but we need a vimeo account to watch the video... from France I don't know for other country

https://waterpixels.net/forums/topic/2654-youtube-quality-issue/#findComment-17905

Thanks for taking the time to do this and share this write up Jero!

Grading is always very subjective, but yes, on the tests I really find the R5II end results the most pleasing as well.

When working in ambient light, I’ve always been impressed by Canon colour science in general - my untested theory is that it while might not be the most accurate, Canon's handling of custom wb allows you to get a colour palette that is aesthetically very pleasing for UW work, something about the way it handles warm hues while preserving blues/cyans, if that makes sense...
Actual colour accuracy would need to be tested in sterile conditions with a macbeth chart or similar, but results ("Canon colours") are really quite magical.

Your conclusions on the RED Helium are very interesting as well.
Brings to mind pics I've seen of rigs used by bluechip productions (BBC crew BTS footage for instance), where operators work mostly in ambient light or filtered mixed lighting to get as close to ambient light as possible. I think they still use a filter on the camera lens in quite a few ambient situations on cinema cameras.

This seems a little anachronistic since you lose stops of light, and the colour science + grading possibilities are massive these days, but could also align with the struggles you mentioned on the RED.
It would be really interesting to compare bare results with results obtained with manual wb + a good filter like the Keldan SF-2, see if this helps the camera adjust for manual WB or not really, especially at good ambient depth of 10m or so.

Really interesting technical note on codec choices as well. I imagine broadcasters usually require ungraded footage to be worked on in-house, hence the need for a workable dynamic range and colour depth.

This also reminds me that as amazing as the footage is on recent BBC underwater documentaries (Blue Planet etc), I’m really not a great fan of the super saturated, often exotic grading colourists seem to come up with these days (which I believe is linked to streaming and phones as primary display devices, attention grabbing all that…).
It's sometimes a little strange to see footage shot in ambient light or aiming to recreate ambient light conditions with filtered Keldans be given this kind of treatment. But maybe I'm just old ˆˆ

As a sidenote, I could watch the clips on vimeo without being logged in to my account for some reason, seems to work from Japan.

And it’s great to see mantas are still around at La Reina!

I was working in LaPaz in 2018 when the mantas came back and stayed, after what had been only very limited sightings for a long while…
Only time in my life I got to see sealions and mantas swimming together, even with a juvenile whaleshark join in the fun once.

Golden memories, and I’m happy they decided to stick around apparently!

cheers
ben

Edited 21 hours ago21 hr by bghazzal

Hey Davide,

Thanks for reviewing my inital findings and making such precise and accurate observations. Much appreciated.

1. Custom white balance at depth
   Yes—I slate with an white balance card every time, nothing fancy. I’ve also bought a neutral‑grey slate (N8) and will test whether it gives more accurate midtones versus pure white.

2. Noise on the R5 II
   I shot S‑RAW in C‑Log 2 at ISO 800 but unfrotunatley underexposed by about 1-2 stops, which amplified grain. Applying ETTR (Expose To The Right) by overexposing log footage by ~+1.5 stops cuts noise dramatically. My exterior tests confirm this aligns with Canon’s and third‑party tutorials on noise management in log.

3. RED color science
   I didn’t use any RED LUTs. Instead, in DaVinci Resolve I applied a Color Space Transform (Log3G10 → Rec.709), then eyeballed contrast and white balance to match Canon’s C‑Log2 Rec.709.

4. 1DX II codec robustness
   The 1DX II’s Motion JPEG intra‑frame codec (8‑bit) remains surprisingly forgiving if exposed correctly. I’ve run it since 2016 as a reliable B‑cam for quick 4K workflows. The R5 M2’s 4K Fine (C‑Log 2) image feels very similar, though it tops out at 30 fps in 4K rather than 60 fps.
   

   Anyway, for me, color has become the most important factor when choosing a setup, especially for dives below 10 meters. Above that depth, most setups will do the job, the differences are minimal. But once you go deeper, color becomes the biggest challenge—at least for me—and as you said, it's very subjective (should we present the subjects in their real color or the color your eyes see at that depth?). That said, color is now my primary tool for creating drama or drawing focus within a shot. I’ve always dreamed of expressing underwater scenes using the chiaroscuro approach to color and light, inspired by artists like Rembrandt, Vermeer, Velazquez and Caravaggio. I don’t know if I’ll ever fully achieve that level of expression with underwater footage, but that artistic ambition keeps pushing me to experiment, learn, and refine every dive/edit.

Let me know if you’d like any deeper dives on these topics.

Hey Ben,

Thanks for taking the time to read and reply to my post—I really appreciate it, and thanks for suggesting I share it here. I totally agree on Canon’s in‑camera color—it isn’t billboard‑accurate, but it’s the most pleasing out of the box.

I hadn’t realized major productions still use in‑housing filters on Nauticam housings to tame their color cast; I’ll look into that. In our tests, my friend operating the camera set a manual white balance on the Kelvin scale—I’ll confirm the exact value. It feels a bit retro to stack glass in front of a modern sensor. I did the same with my Sony rigs from 2005 to 2009, but in 2025 it seems counter‑intuitive.

Regarding BBC‑style saturation (Ocean, Blue Planets), I believe much of that punch comes from shooting on RED sensors with their extended latitude and then aggressively grading in post. I have no issue with their surface shots—those baitball colors are superb. However, at depth they’re effectively “bringing back” reds filtered out by water, and I don’t see how this can be done effectively on a RED setup, especially after working with the files from our recent dives in Cabo Pulmo.

I’m a big fan of vibrant color and dramatic chiaroscuro underwater—nothing beats a rich, directional key light against inky shadows. To avoid a forced look, I rely heavily on the scopes in DaVinci Resolve. For example, here’s how I use the vectorscope’s 2× zoom to nail saturation without clipping:

Using the Vectorscope 2× Zoom in DaVinci Resolve

1. Open the Scopes panel (Workspace → Scopes).

2. Click the Vectorscope icon, then click the magnifier twice to engage 2× zoom. This enlarges the color plot so you can see individual saturation vectors more clearly.

3. Enable the “Skin Tone Line” overlay (right‑click the vectorscope → Show Skin Tone Line).

4. While adjusting the Hue vs. Hue and Hue vs. Sat curves, watch that your color points stay within the broadcast‑legal circle and align with the skin‑tone line when balancing midtones.

5. Use the zoomed‑in boundary as a visual “fence” to prevent oversaturation—if your vectors push past the circle, dial back your curves or lower the Chroma Gain.