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

List:       koffice-devel
Subject:    Re: GSoC 2009 proposal, looking for mentor
From:       Boudewijn Rempt <boud () valdyas ! org>
Date:       2009-02-25 10:17:11
Message-ID: 200902251117.12094.boud () valdyas ! org
[Download RAW message or body]

On Tuesday 24 February 2009, LukasT.dev@gmail.com wrote:

> I'm also looking for mentor. I talked with boud and Cyrille and they both
> seems to not to mentor this year. So who is going to mentor for KOffice?

I'm still not completely decided, but I would like to have a bit of a breather 
this year, esp. since last year I took two students. But: we really need 
mentors to step up! (Also for other apps.)

> 1. 3D model of brush
>
> Paintops are sometimes also modelled as 3D device. I gathered some
> experiences with 3D last semester and I would managed to do some 3D
> model.
>
> I already gathered some publications about various brushes (even 3D)
> and Baxter work also contains useful informations too.
> http://wiki.koffice.org/index.php?title=User:LukasT
> I also received some GPL2 code from a student who did 3D brush as
> thesis and it is waiting to be ported to Krita
>
> Excepted results: paintop that is based on 3D model and has stronger
> physical attributes than current brush engines. Bidirectional painting
> support.
>
> Something like this in short http://youtube.com/watch?v=W4suJCMdnsU
>  and this http://youtube.com/watch?v=CNfAXjRQkr0
>
> boud and Cyrille: +1

Yeah, this would be nice. Though finishing the painterly framework with 
features like wetness etc. is important, too.

> 2. Diffusion curves
> http://artis.imag.fr/Publications/2008/OBWBTS08/
> I already emailed authors:
> Main problems:
> * solving a Poisson equation AFTER rasterization.
> This implies that:
> 1) The rasterization is not a classical one, so you wouldn't be able to
> represent the images in classical vector formats like PostScript, for
> example.
> 2) The heat diffusion is quite slow when solving it on the CPU, so we used
> the GPU. But this restricts the use of Diffusion Curves to people who
> have good GPUs.
> 3) We can solve Poisson equation on big images, but not on infinite ones.
>
> Excepted results: New tool - diffusional curves in Krita/Karbon or
> both (don't know about possibilities of sharing this tool)
>
> Cyrille added:
> Deffinitively for Karbon, and also very cool. The point 2) isn't really a
> problem, if you use the correct programmation language, I invite you to
> look at OpenCL for this, it should be available in Nvidia and ATI
> proprietary drivers, shortly. And it will come as open source in the next
> two or three years. The advantage of OpenCL is tht it can run on GPU or
> CPU.

I haven't got a clue about this one. Consequently, it doesn't excite me much.

>
> 3. Advanced interpolation
> There is need for good scaling algorithm for images in Krita and maybe
> in other apps too. Lately I found this discussion interesting
> https://lists.xcf.berkeley.edu/lists/gimp-developer/2009-January/021512.htm
>
> Basically it talks about various improved scaling and interpolating
> algorithms. So the aim of this project would be to do some kind off
> research and implement as much as possible many advanced interpolating
> algorithms so
> that we can compere.
>
> Excepted result: better scaling possibilities of graphics data in KOffice,

I wouldn't spend time on this: we're quite okay now after boemann's latest 
patches and there's much more urgent work to be done.

> 4. Support for animation
> This would be quite hard. Make 2D renderer. MNG or own file format for
> saving, export every frame as [PUT FILEFORMAT HERE]
> I had idea in which every layer is a scene and for start few things
> could be animated:
> * Basic motion
> * Filters and it's values for the whole layer
> * Paintops and their movements
> * Geometric transformations +Warping
> I would propose some kind of system that Blender uses and in Blender
> it is call IPO curves.
> I don't know if this would be for Krita or Karbon.
>
> Inspiration: Blender, http://www.les-stooges.org/pascal/pencil/ ,Gimp
> Animation Pack GAP,
>
> Excepted results: Timeline, animate able KisObjects, animation support
> for filters and import/export filters for it. It should not just flip
> layers.
>
> Cyrille:
> While I love that idea, and would like to see it happening in the future
> (possibely with the upcoming Qt's animation framework), including animated
> svg, I fear this is too huge for a google summer of code project.

Yeah, pretty huge. There's also overlap with KPresenter, which should get a 
complete SMIL implementation. Having SMIL, and krita or karbon images, one 
could argue that there is a complete animation solution just waiting for a 
nice user interface. But this is a big job, a really big jog.

-- 
Boudewijn Rempt | http://www.valdyas.org
_______________________________________________
koffice-devel mailing list
koffice-devel@kde.org
https://mail.kde.org/mailman/listinfo/koffice-devel
[prev in list] [next in list] [prev in thread] [next in thread]