Re: opengl and images (was: Re: fltk backend fix & mouse wheel scroll factor)

[Shai Ayal]

On Sat, Oct 3, 2009 at 4:48 PM, John W. Eaton <address@hidden> wrote:
> On  3-Oct-2009, Shai Ayal wrote:
>
> | On Thu, Oct 1, 2009 at 9:21 PM, John W. Eaton <address@hidden> wrote:
> | >  image (rand (8, 10, 3))
> | >  set (gca, 'xlim', [2, 4])
> | >
> | > I would expect something like the first image attached below
> | > (generated with the gnuplot backend), but I get the second.  Oddly,
> | > the axis is resized to fill the screen if the window is refreshed (by
> | > placing another window on top of it then uncovering it, for example).
> | > I'm not sure what the proper fix is.  Any clues?
> |
> | I attach a changeset to fix these issues -- It clips the image
> | according to axes limits. It also fixes the limits of the image object
> | to take into account the real size of the image
>
> I applied the patch, but it still does not seem to behave as I would
> expect.  Now I see the following after doing
>
>  image (rand (8, 10, 3))
>  set (gca, 'xlim', [2, 4])
>
> (first attached figure) when I would expect something more like the
> gnuplot output (second attached figure).
>
> Do you see a way to display the  partial pixels between x=[2,2.5] and
> x=[3.5,4]?
No. This a limitation of OpenGL images -- no partial pixels. Do you
think this is serious enough to warrant a rewrite in terms of another
OpenGL primitive?

> Also, do you understand what is causing the odd scaling of
> the axes?  they are adjusted if the figure window is redrawn (for
> example, but covering it with another window and then uncovering it)?
I don't get it either.

Shai