Interesting development for underwater imagery / photography

[Divingblueberry]

A very interesting innovation :

I find the title of the video a bit misleading, but the results are impressive.


For those interested in the exact science, the publication is here:
http://openaccess.thecvf.com/content_CVPR_2019/papers/Akkaynak_Sea-Thru_A_Method_for_Removing_Water_From_Underwater_Images_CVPR_2019_paper.pdf

Safe diving

 

[Fastmarc]

Interesting indeed.

I'm a bit torn as to whether I would want the scene to look as it was on land though. I want it to look as how I would be seeing it if diving and that's what I would want to share. I think that would be my personal preference.

However, this will be great for other applications for sure.

 

[Steve_C]

Interesting. Does seem to need fairly clear water. Not clear that the effect of distance is fully removed. Things still looked very hazy in the distance. But for close to intermediate it helped a lot.

 

[wetb4igetinthewater]

Very interesting. I'm curious as to how this technology gets applied/utilized in the future.

 

[aviator8]

interesting. I wonder if she is targeting to only use this in post or apply the algo in camera software. She needs a smaller wing too.

 

[wetb4igetinthewater]

She needs a smaller wing too.

And get rid of those split fins!

 

[Mo2vation]

Holy Taco Wing, batman....

Yeah - What do you expect when you dive with no strobes...? Also, who's diving with an DSLR in an Pro-level housing shooting WA with no strobes?

I like this technology... a lot. When I shoot, the whole idea of Wide Angle is to make the water disappear and focus on the subject(s). You can generally make the water disappear with proper lighting and settings. Albeit in a much smaller scale, more intimate scale.

Dig it. Truly. Thanks for sharing.

Not sure we'll see this on a retail site anytime soon, as it sounds like its geared towards institutional and scientific use.

Someone will buy her numbers, build it into a filter and you'll see it on Facebook soon for all the GoPro folks. That's where she'll get her cash back. Selling to GoPro'ers.

- K

 

[seeker242]

Just came across this today. Very interesting!


Some comments from the author.

"Just a clarification: the method does NOT require the use of a color chart. That part was not clear in the video. All you need is multiple images of the scene under natural light"

"working on implementation as an app, or in-camera now. Stay tuned."

"It will be available free for academic/non-commercial purposes very soon, commercial use will be later next year"

The actual science behind it.

http://openaccess.thecvf.com/conten...er_From_Underwater_Images_CVPR_2019_paper.pdf

 

[Jiminy]

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

 

[WreckDiver1692]

Found the redditor

This is cool I just hope it becomes available at some point, sooner rather than later. The underwater color science of Nikon and Canon for their high-end cameras is pretty good now but this could probably make it even better. Whether this is something implemented in-camera or occurs solely in post I suppose depends on the complexity / computational overhead of the algorithm.