Interesting development for underwater imagery / photography

[Mo2vation]

I don't see anything revolutionary. It looks like she is just using a standard color target under water.
If you shoot in raw and and use a raw processor that works with a reference target like the one below you should be able to do the same.

Colorchecker Classic - Camera & Image Calibration: X-Rite Photo & Video

Yup!

Or, if you're using proper strobes, WB on something on your subject. My order of preference:

1) Suunto Compass Bezel (nearly perfect neutral gray)
2) Atomic Reg Trim (also nearly perfect neutral gray)
3) Scooter Zinc
4) Diving Steel Cylinder

Her whole thing is color without strobes. She needs to get to the GoPro market for mo' money.

- K

 

[Brian G]

The algorithm depends on distance information for each rgb "pixel". They said they got that information via a "Structure From Motion" algorithm, meaning they did video processing that generates relative distance and calibrated it with objects of a known distance. They mentioned that you could also get that info processing still images from a stereo camera. In short, it will take a while before this becomes easily accessible.

 

[Divingblueberry]

And get rid of those split fins!

LOL ! ! !

I wasn't expecting that one

 

[Doc Harry]

I had been doing this in Photoshop for years. It's a matter of replacing the lost reds using layers.

 

[wstorms]

Isn't this the same the Paralenz camera does with video? Apply a depth based colour filter?

 

[Schwob]

I had been doing this in Photoshop for years. It's a matter of replacing the lost reds using layers.

How does one replace lost reds? I fathom boosting subdued reds (and am not good at it), but when there is no red Info in a pixel because of depth / distance to subject, then how does one replace that proportionally correctly?

And I am not saying her technique does that. At shallow depth it maybe could due to using info from various distances in the computation. Real deep, there‘d have to be a light brought to those multiple distances ... I‘d think... And then, I am probably not understanding it at all...

So is her technique about getting the color right or at the same time also about getting all the floaters in the entire water column visible in the picture out as well? Haven‘t eve digested that yet on a fact basis, never mind truly understanding how...

 

[Hatul]

There is an iOS app called Dive+ that does a great job of color correcting my underwater photos in my iPad photo library, but I've not figured out how to use it to correct my GoPro videos.

 

[Laurie S.]

I'll stick with my strobes...and split fins.

 

[doctormike]

When I first read about this, I thought - wow, she invented white balancing. But it seems that there is more to it than that, there is some pretty complex math and backscatter reduction, etc...

It sounds as if it is designed to help computer image processing systems do scientific research without the need for human operators to manually scan each image. Not really meant as a replacement for strobes or for depth balancing filters.

Here's the original paper, if you want to learn more than just what has been circulating in the press...

 

[tursiops]

She is trying to solve the first-principles forward problem...given the properties of the sea water, how to remove its effects from the image? The properties are ascertained by multiple images at different distances.
Another way of doing this (what we normally do) is the inverse problem, i.e. to assume something about the image -- what is "white" for example -- and then forcing that to be white in the image.
I don't see what all the fancy math and calculations provide that we can't already do -- with just one image -- and we don't have to have a static image that we must look at from various distances.
What she calls backscatter is not what underwater photographers call backscatter; she is removing the accumulated effect of minuscule particles in the water scattering light of different colors...we refer to single, larger particles (but often very many) scattering light as points of light...and it is not usually a problem with ambient light photos anyway.