
How to edit your underwater photos and videos? The ethical AquaExposure method: restoring colours stolen by water, not fabricating a reality that doesn't exist.
To learn how to get the most out of your gear underwater, discover the AquaExposure training.
I remember my first dive in the Maldives, at 18 metres, with my Nikon D810. I'd filmed an absolutely spectacular manta ray passing two metres from me. Back home, I looked at the photos for real - everything was blue-grey, as if I'd just filmed an underwater horror movie. The ray had the same depressing tones as my wool jumper.
I panicked. "I've ruined my photos. I'm going to quit underwater photography."
That's when I understood something fundamental: what the ocean had stolen from me wasn't my image quality. It was the colours.
Ansel Adams, the legendary nature photographer, said it better than I could: the negative is the composer's score, and the print is the performance. In other words, what you capture is only the first half of the work. Editing isn't cheating - it's creation.
Except there's an enormous trap.
The scientific figures: - Water absorbs red/orange wavelengths from 5 metres - At 10 metres, you've lost 90% of reds - At 20 metres, it's over 99% of reds AND greens - What remains is primarily blue and cyan
This isn't a "bad technique" issue - it's physics. Water absorbs photons starting with the longest (reds) and finishing with the shortest (blues). It's predictable, quantifiable, and completely reversible.
Laurent Ballesta, the Franco-Monegasque photographer and diver who documented the living fossil Coelacanth, says something important in his interviews: ethical editing means giving back to the image what the medium stole from it, not inventing a reality that doesn't exist.
There's the limit. And it's clear.
Restoration does not equal Artifice.
Restoration: you return the colours the fish TRULY had. Artifice: you invent a texture, a tint, an atmosphere that didn't exist in the ocean.
Before touching a single slider, you need to isolate your subject. Why? Because restoring the colour of the subject AND the background simultaneously creates a "fake" image - the underwater background can't have the same saturation as your subject.
In Lightroom: - Tools > Linear or Radial Correction - Select your main subject (the fish, the coral) - You now isolate the edit to this zone
In DaVinci Resolve (for video): - Fusion > Node - Use tracking tools to follow your subject through the video - Apply the correction only to this zone
Let me tell you about a blunder: I spent a long time applying corrections to the entire image. Result? My fish looked overexposed and the backgrounds resembled a post-war photograph. That wasn't restoration. That was massacre.
Now? I take the time to select properly. Yes, it's 2 extra minutes per photo. But the result is a thousand times better.
Once your subject is isolated, you'll restore the vanished wavelengths.
The simple rule: - Increase red/orange saturation first (they were the first stolen) - Then greens - Leave blues alone - Control the overall colour temperature
In Lightroom: 1. Go to the "HSL" tab (Hue, Saturation, Luminance) 2. Click on "Saturation" 3. Increase REDS by +30 to +50 (start moderately) 4. Increase ORANGES by +20 to +40 5. Increase YELLOWS by +10 to +25 (less aggressively) 6. Don't touch the blues unless you have a specific problem
The classic mistake: Increasing the overall image saturation. No. You'll make the blues overexposed and the image looks "fake" immediately.
In DaVinci Resolve (for video): 1. Go to Color > Curves 2. Use the separate RGB curves 3. Increase the RED curve in particular 4. Increase the GREEN curve slightly 5. Keep the BLUE curve stable or slightly reduced
Result: your manta ray, which looked depressingly grey-blue, becomes GREEN. Yes, really green. Because it WAS green.
This is where most people trip up. You've edited your image on a screen. But how do you know your screen is displaying colours correctly?
Screen calibration: Not glamorous, but crucial. A Datacolor Spyder or X-Rite is an investment of 100-200 euros that saves you hundreds of failed images.
Export for Web vs Print: - Web: sRGB, 8-bit is sufficient - Print: ProPhoto RGB, 16-bit minimum
Here's a tip: before printing your photos, test on a small 10x15 print. Just one. Pay 3 euros for a test photo. It'll save you the humiliation of ordering 30 poorly coloured prints.
Video is more complex because every frame needs to be consistent with the next. If colour jumps between two frames, it screams that there's been editing. And it's when it's badly done that you notice.
DaVinci has a "Stabilizer" tool in the "Fusion" tab > Camera Stabilizer. It's free. It's effective.
How to: 1. Drop your clip on the timeline 2. Go to Fusion > Stabilizer 3. Analyse (it'll take a few seconds) 4. Apply with 50-80% strength (too strong = distortion)
Why a single Node? If you correct each clip individually, you'll get unpleasant colour "jumps" between transitions. Your brain will immediately see: "cut-cut - oh, the colour changed".
The method: 1. Place all your clips on the timeline 2. Go to the "Color" tab 3. Create a Node (cmd+D on Mac, ctrl+D on Windows) 4. Apply an identical correction curve to all clips
The ideal curve: - RED curve: slightly increase the midtone (+15-25%) - GREEN curve: also increase the midtone (+10-15%) - BLUE curve: keep stable or reduce very slightly (-5% max)
Result: chromatic consistency across your entire video. Colours no longer "jump".
How to isolate in DaVinci: 1. Create a second Node 2. Go to the Qualifier or Mask tab 3. Select the zone/colour you want to correct 4. Apply your correction (HSL as in Lightroom)
It's more advanced than Lightroom, but the concept is the same: isolate to control.
This is the big revolution of the last three years. AI upscaling tools can transform HD video (1920x1080) into "4K-like" quality (3840x2160). Topaz Gigapixel AI or even open-source tools like Real-ESRGAN do genuine work.
The numbers: - Real-ESRGAN 4x: upscale x4 resolution (1080p to apparent 4K) - Topaz Gigapixel: upscale x2-x4, simpler interface - Cost: Real-ESRGAN is free, Topaz around 80 euros
But beware: AI invents details that didn't exist. This isn't "restoration" - it's augmentation. If you sell upscaled content claiming "native 4K", you're lying.
My advice: only use AI upscaling if your video is genuinely in HD. If you're already in 4K, leave it alone. It's a waste of CPU.
Lightroom presets for underwater photography? Very popular. Very dangerous. And I'll explain why.
Every dive has radically different water conditions.
Maldives at 10 metres: Clear water, lots of blue light, moderate red/orange restoration needed.
Maldives at 25 metres: Very dark water, very little red light, MASSIVE correction needed. Reds must be doubled.
Caribbean at 15 metres: Yellow/green water (Jerlov Type II), completely different correction. Blues are less dominant, greens return on their own.
Scottish loch at 8 metres: Brown water (tannins), no clear blues at all. You correct as if you had a brown cast.
A preset that works for ONE photo will NEVER work for another. You'll need to adjust it, modify it, fiddle with it. And at that point, you've done just as much work as if you'd learned HSL.
My firm recommendation: Forget presets. Learn HSL principles.
Spend 30 minutes understanding how to adjust reds/greens/blues, and you'll be able to correctly edit any underwater photo without wasting time on subsequent adjustments. It's a time investment that pays immediate dividends.
(Concept: include 3-4 before/after pairs showing the progression)
Example 1: Manta Ray in the Maldives (18m) - Before: Grey-blue, no texture definition, looks like a ghost - Correction applied: +40 reds, +35 oranges, -20 blues - After: Natural colour (grey-blue-white belly), visible body texture, expressive eyes
Example 2: Black Coral on video (20m) - Before: Completely invisible black in the frame, looks like a void - Correction applied: Red curve +50%, secondary node isolated on coral only - After: Coral visible with detailed texture, branching visible, dark red colour visible
Example 3: Clownfish in 4K (12m) - Before: Orange too faded, facial features invisible, eyes closed - Correction applied: +50 oranges, +25 reds, isolated selection on the fish - After: Vibrant orange, sharp and expressive eyes, black patterns visible
Example 4: Caribbean fish school (15m) - Before: Opaque yellow-green, all fish look the same, no contrast - Correction applied: Moderate red increase (+25), green increase (+15), blue reduction (-10) - After: Each fish has its own specific tint (some yellow, some red), contrast between them visible
These examples show a reality: corrections are NEVER the same. You need to learn the principles, not memorise numbers.
No. It's restoration. You give back to the image what water stole. It's ethical as long as you're not creating a fish that didn't exist or a coral that wasn't that colour.
Simple test: Show your edited photo to a diver who was with you. They should recognise the scene and say "yes, that's what I saw, though it wasn't quite that colourful in reality."
Lightroom for photos: the simplest and most powerful tool. DaVinci Resolve for video: free, professional, with advanced colour correction. Capture One if you have the budget: simpler interface, slightly better results.
Go to Lightroom > HSL > Saturation > reduce BLUES by -15 to -25. Then increase reds/oranges. That's it.
Edit manually. Presets are traps. Ten minutes of manual editing beats an hour of fiddling with a preset.
DaVinci Resolve, Color tab, single Node for the entire sequence. Use the RGB curves to increase reds/greens. It's the standard tool in video production.
AI upscaling is a process where artificial intelligence "invents" pixels to enlarge an HD image to 4K. It's ethical if you mention it (it's not true 4K). It's not ethical if you sell content presenting it as "native 4K".
Yes, if you extract from 4K video. A frame extracted from 4K video has exactly the same resolution as a 4K photo (3840x2160). Quality remains identical.
No. Read more by clicking "Photo vs Underwater Video".
I know a photographer who edited his underwater photos so much they look like 3D renders. Saturation at +100. Extreme contrast. Blue halos around fish.
That's not restoration. That's visual lying.
The golden rule: If you look at your edited photo and think "wow, that can't possibly be real", you've over-edited.
Good editing is invisible. The viewer looks at the photo and says "wow, that fish really looked like that" - not "wow, what incredible editing technique".
Take your best underwater photo (the one where you thought "shame it's so blue").
If you find it better, you've just understood the principles. Congratulations, you're officially an ethical underwater editor.
If you find it over-saturated, halve the values. That's normal.
I wrote it sitting in my office in Brussels, thinking about all the photos I ruined before understanding that editing wasn't cheating. Every screenshot comes from my real sequences, every piece of advice comes from a struggle I've lived through.
If you have questions: [email protected]
Benjamin Coste Founder, AquaExposure
For any questions about underwater editing, join our AquaExposure community.
You can correct what the water distorted, colours, contrast, white balance, without inventing what was not there. The limit is staying true to the scene you witnessed.
No. Water absorbs colours with depth, so restoring red or blue brings the image closer to reality, not further. Cheating would be adding an animal or a backdrop.
Not necessarily. Excellent results are possible on mobile with Lightroom or Snapseed. Skill matters more than the price of the tool.
Reveal without betraying. You enhance the natural light captured at the shutter, you do not fabricate a scene. The best edit starts underwater, with a good base image.
Yes. You correct colour drift and contrast, you stabilise, but you keep the real mood of the dive. The honesty of the document comes before the effect.