[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-kimageshop
Subject:    Re: Q. on Colorization Plugin
From:       "Schleimer, Ben" <bensch128 () yahoo ! com>
Date:       2007-01-05 0:33:38
Message-ID: 402881.6944.qm () web32009 ! mail ! mud ! yahoo ! com
[Download RAW message or body]

Hi Bala,
   I can't tell what the details of SIOX are (because
they didn;t publish the algorithm on their site) but
the re-colorization algorithm doesn't require any
alpha. Basically, the image is converted into YUV
colorspace. The Y channel is always kept but the U and
V channels propogate depending on how similar the Y
channel is of ajacent pixels. Most of the repainted
examples in the paper are grayscale so I'm not sure
you could emulate SIOX 100%. I guess if you paint a
region a funky color and then generate a colormask in
YUV space and then apply that mask to the RGB image,
it might work... however IANAE

Cheers
Ben 

--- balachandran c <balachandran_c@rediffmail.com>
wrote:

> Hi Ben,
> 
> >developing krita plugins (specifically, the
> >re-colorization algorithm from
>
>http://www.cs.huji.ac.il/~yweiss/Colorization/colorization-siggraph04.pdf)
> 
> In the paper, it says that you color samples of
> pixels from each region with desired paint. The
> color is then "propagated" along similar values of
> L. (In a manner of speaking). If I paint an object
> with say alpha = fully_selected, and the outside
> region with alpha = 0, will the algorithm result in
> picking out the desired object, like SIOX tool does?
> Or can it do that with trivial modifications?
> Multiple birds, single stone etc..
> 
> Cheers,
> Bala.
> 
> 
> > _______________________________________________
> kimageshop mailing list
> kimageshop@kde.org
> https://mail.kde.org/mailman/listinfo/kimageshop
> 

_______________________________________________
kimageshop mailing list
kimageshop@kde.org
https://mail.kde.org/mailman/listinfo/kimageshop
[prev in list] [next in list] [prev in thread] [next in thread]