[vlc-devel] Re: Question about video filters - svg rendering

[Derk-Jan Hartman]

I just saw this question was still unanswered, and since i personally 
like SVG a bit, i though i'd answer it anyway and ask if you have made 
some progress.

DJ

On 06 jan 2005, at 15:59, Olivier Aubert wrote:
> Hello
>
> I am trying to update the SVG rendering module that worked in 0.7.2 to
> use the new filter API of 0.8.1. It should be much simpler now that
> there is a blending infrastructure, but I think I did not quite catched
> how it works.
>
> What I do is basically (using freetype.c as model) :
> - in the RenderText function, render the SVG data into a GdkPixbuf that
> is stored in the associated subpicture_data_t

Why is this stored in a GdkPixBuf?
seems kinda unnecessary to me. only makes it depend on GDK. can't it be 
copied directly into a picture_t ?

> - in the Render function, copy the rendered GdkPixbuf to a new
> subpicture region
>
> It does not work (displays nothing), so here are my questions :
>
> - I assume that the Render method is passed a subpicture that is meant
> to be filled with regions holding supplementary data. The additional
> regions are to be rendered into the final picture by the VLC
> infrastructure. Am I deeply wrong ?


> In the previous version (the one that is still in svn:
> module/misc/svg.c), I had to write the merging code myself, as I was
> given a picture_t. I hoped (from what I saw) that the changes removed
> this burden from the modules to the infrastructure. Here, a GdkPixbuf
> happens to be a simple RGBA buffer, so I figured that I would simply
> create a new region with a FOURCC RV32 and copy the pixbuf data to the
> region's first plane.

Our core can only handle YUVP or YUVA blending. so your model needs to 
output a picture in this format.
these chroma formats are really quite simple, they just have an alpha 
plane next to the YUV planes.

> - Is there any way for the filter_t* object to access the associated
> video_output_t properties ? I need the width and height of the
> video_output, as parameters for SVG rendering (and having matching
> dimensions should speed the merging process).

It's not easily posible i think, but gibalou/Gildas probably knows.

DJ

---
Universiteit Twente
Derk-Jan Hartman (d.hartman at student.utwente dot nl)
http://home.student.utwente.nl/d.hartman

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html