www.ssontech.com

[joel_lelievre]

Here's one for ya... I have a shot that takes place underwater that I need to track. For the most part it's a simple camera pan from left to right with some slight motion up and down. After doing some research here on the forums and reading over the Pathology document I have learned that it can be quite difficult to get a 3d solve on a shot that involves a pan due to lack of perspective change. So that grouped with the fact that it takes place underwater with bubbles and debris floating around, and the light changes due to caustic reflections, I am having a bit of trouble attaining a nice track. Just wondering if anyone has worked with underwater shots in the past and if so what advice could they lend.

Here is what I am up against:


thanks in advance

[kenekeu]

Coming from an image manipulation background I would suggest running the sequence through a color correction with a midtone lightening so you can get details on the ground. Go for your solve on the lightened clip then apply to the darker. Just a thought. Anyone else...

[gordonrobb]

I am only speculating here, and I'm sure Russ will let you know, but if you are looking for a 3D solve wouldn't the refraction caused by the water screw up any calculations?

[ssontech]

The refraction difference should only affect the effective focal length/field of view of the lens.

[Smola Martin]

Question is ... What will want to do with shot after? Will past other CG object or what .........?

Would you mind send part shot. Picture is nice but here isn´t action.

[joel_lelievre]

Thanks for the replies everyone. Here is a link to the raw footage I am working with. I have played around with it and can get some semblance of a track but it still requires work. I'll try to post the actual tracked version later on.

http://www.intrinsia.net/transfers/underwater.mov

[Smola Martin]

it´s really crazy shoot for tracking -)
... in my opinien --- here isn´t camera move .. only rotate (it mean you can´t solve camera trajectory because here isn´t) In future If you want camera trajectory you must to swim around or along object and film.

.... you can compute tripod solution or for example oceanaut rod move

.... have you any information about camera and set up camera this shot?

.. I try several track for tripod. Here are many move objects and reflex. All must be hand tracking here is ocean rock. But I am not content wit result. At the weekend I will try it again. I try use different track ... window track maybe wasn´t good choose.

Here is short preview very huge value - distortion -0.03205 and FOV 63

[andronikos916]

My opinion for that shoot is:

- Find a decent Lens Distortion Remove setting
- Manual tracking
- Tripod Mode
- Apply Lens Distort

Thank you,
Andronikos