On 02.06.2015 22:36, Dan Dennedy wrote: > On Tue, Jun 2, 2015 at 11:05 AM Brian Matherly <c...@brianmatherly.com > <mailto:c...@brianmatherly.com>> wrote: > > You make a good point. The same result can be achieved by applying > the affine transition. Instead of putting the rotation > responsibility on the pango producer, why not allow the watermark > filter to use the affine transition? > > > Umm, no. melt +hello.txt -attach affine transition.fix_rotate_x=45 > > filter_watermark is to transition_composite as filter_affine is to > transition_affine. I do not want to extend watermark to use affine when > filter_affine is so similar: > melt +hello.txt -attach affine:noise: transition.fix_rotate_x=45 > > Neither will I accept adding any more features to the already unwieldy > transition_composite. One can use tracks and transitions - possibly > encapsulating it within a tractor as a virtual clip - either inline or > as distinct MLT XML. Also, one can work to improve affine or > core/filter_transition. > ...assuming I have decided to reject the idea of rotation within pango. > Let me hear Maksym's reaction/argument.
my arguments are: - *quality*: basically glyphs rotated at vector level and only then rasterized subpixelly - *CPU usage*: it does not perform any floating point calculation and image scaling. it is basically pango renderer feature like pixbuf scaler in producer_pixbuf, why does not use it? -- Maksym Veremeyenko ------------------------------------------------------------------------------ _______________________________________________ Mlt-devel mailing list Mlt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mlt-devel
