[Bf-committers] GSoC 2016: Finish multi-camera reconstruction

[Sergey Sharybin]

Hi,

Answers are inlined.

On Wed, Mar 2, 2016 at 6:40 PM, Matthias Fauconneau <
matthias.fauconneau at gmail.com> wrote:

> Hi Sergey,
>
> I was thinking of participating to GSoC again this year.
>

That's good!


> If we could allocate two slots on motion tracking, we could do an UI
> / backend cooperation like I did with you in 2011.
>

Guess you're talking about collaborating with Tianwei Shen on this project?
Then it's a bit tricky to separate scopes too much at the current state.
Additionally, i don't see just working on interface for a new tools is a
sufficient GSoC project. It's not _that_ much work in UI to be done anyway.


> This would allow us to expand the scope of the project to make
> sure Blender meets users requirements for VFX.
> The other two smaller projects could be folded in since:
> - Variable camera lens focal length support should be a matter
> of integrating the additional parameter in the resolution.
>

That is indeed a nice project.


> - Lens distortion estimation and grid calibration backend was done in 2011
> and is mostly frontend work at this point.
>

It's not just a frontend work. The project was just a hook to OpenCV, which
is not acceptable for Blender's upstream. Additionally, we now can do
much-much better job on estimation/calibration using Ceres.

-- 
With best regards, Sergey Sharybin