(Started new email so title is more relevant). Included in today's GIT checkin is the following. IgorB supplied another 25 Shape Wipe Transitions -- actually 50 with a 1080 and a 540 version. For further information refer to earlier email at: https://www.mail-archive.com/cin@lists.cinelerra-gg.org/msg00619.html GG did a lot of rework on the Shape Wipe Transitions code. He added the 25 1080 versions; fixed a couple of bugs; added a Feather slider (based on exponential values); and built in parallel processing instead of single threading. So in response to:
Yes it is resampled, but by my tests on my poor PC, 540 works faster than 1080 for a 29.97 fps project, the MIN fps achieved with "Tr-1080_Butterfly" is 22fps, while with "Tr-540_Butterfly" is 28fps. So, in some projects I can use the 540 instead of 1080 to save cpu.
the fps with the 1080 should now be almost the same as the 540 now with the rework that gg did. I tested it and it seemed to use the same amount of cpu at 1080 as at 540 (monitor with the "top" command and checked Preferences frames per second). He says that once past the initial check at the very beginning, it should run at the same speed.. A static tar build is at the following in case anyone wants to test for Ubuntu16: https:// cinelerra-gg.org/download/testing/cinelerra-5.1-ub16-x86_64-static.txz Thanks IgorB! gg/Phyllis