[CREATE] ORA fullscreen viewers

[Jon Nordby]

On 9 September 2010 23:41, Martin Renold <martinxyz at gmx.ch> wrote:
> We want to make the image viewer's job easy and reliable, or else they will
> hardly bother to add support.  ORA fallbacks are still not specified, making
> it impossible to implement a full-resolution ORA renderer that will not
> screw up in the future.
>
> I don't expect an ORA viewer to implement many blending modes (if at all),
> nor to render a tiled pattern.  But would it be asking too much to do an
> OVER composition?  Is this just a MyPaint specific problem, or is my
> reasoning true for the needs of other programs?
I doubt this is specific for MyPaint, it's just more visible as we
(and our users) actually use OpenRaster to a significant extent.

As I see it there are three feasible solutions to the problem.
1. Have full-image fallbacks in the file, which would then be trivial
to open for any application.
2. Expect viewer applications to implement basic ORA. This requires at
least parsing the basic XML layer stack, and do OVER compositing. In
the future this should probably also include layer-fallbacks.
3. Let a common library implement basic ORA support, and provide an
interface to get the full image as a pixel buffer.

My favorite option is number 3, mainly because:
1. results in redundant information in every new ORA file, as you
point out. And it does not work for existing files.
2. results in fairly significant duplicate work for each
implementation. Or the current state - no implementations at all.

In addition having a library abstract the details away means that
newer features could be added in a transparent way as the standard
develops.

Do note that after MyPaint 0.9 I plan to shift my focus to OpenRaster
issues for a while, so I'm glad you're bringing this up.

-- 
Regards Jon Nordby - www.jonnor.com