Am 28.07.2012 09:47, schrieb William Bader:
If there is an issue of speed (for viewing on the screen) vs. accuracy
(for making a high quality eps), maybe it could be a command line option.
I agree, William. That was also at the back of my mind. But let us have
a reliable better algorithm first. I often see ideas here on the list
but often also without any response or results later :-(
Thomas
William
> Date: Sat, 28 Jul 2012 09:33:38 +0200
> From: [email protected]
> To: [email protected]; [email protected]
> CC: [email protected]
> Subject: [poppler] rotated/downsampled images
>
> Hi Arpit & Bill!
>
> In a quite short period You both offer to implement a better algorithm
> for rotated / scaled images than the Bresenham algorithm, one of You
> wants to implement the bilinear, the other the bicubic interpolation.
> Would be funny to see, which one is the better compromise between speed
> and quality.
> In splash always the scaling of an image is done first with
> Splash::scaleImage, and then the scaled image will be rotated in i.e.
> Splash::arbitraryTransformImage. So this is the place where You can
> implement a new algorithm.
> I know, that we get sometimes a poor quality for slightly rotated
> images, see i.e the wheels in
> http://www.it.piaggio.com/media/mp3hybdepita.pdf on the first page, and
> therfore it would be nice to have a better algorithm. But we must keep
> the speed in mind, because poppler is often used to display a PDF in
> time under unix. If You need any assistance, i.e. proof Your new
> implemented code with a lot of PDF, I'm willing to support You.
>
> Cheers,
> Thomas
>
> _______________________________________________
> poppler mailing list
> [email protected]
> http://lists.freedesktop.org/mailman/listinfo/poppler
_______________________________________________
poppler mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/poppler
