On Sat, 03 Feb 2007 13:22:18 JST [EMAIL PROTECTED] (Andrew Church) wrote: > >There is also one thing that I wish to know for a long time -- how and why > >transcode determines how to auto resize. > > That's something Francesco will have to answer, as he's done all the > profile work. (Assuming you're talking about CVS HEAD; --export_prof > does strange things in 1.0 that sometimes work and sometimes don't.)
So here we go. The situation is simple: there aren't improvements in resize logic for profile code in CVS HEAD. If 1.0.x does strange things (and looks like it does...) when choosing resizing values, almost certainly 1.1.x will do too. Yet. This was deliberate. I don't minded to improve autoresizing since in profile work I done this was not one of my objectives. The main task was to make profile code more modular, tunable and let the user change easily profile presettings (= without to change source code and recompile). Of course, autoresize code *needs* sanitization, but I plan to do this once new profile infrastructure is in place. We're pretty near to this milestone: we need just a few more NMS-powered modules and we can start testing. encode_lavc (and maybe multiplex_lavf) are key factors (since formerly export_ffmpeg embedded ~90% of profile support). I'll very happy to discuss autoresizing strategies/algorhytms here or in -devel. Just spoke up! :) ++ Just for the record, the existing auto resize logic have some dark corners for me too. :) Bests, -- Francesco Romani - Ikitt ['people always complain, no matther what you do'] IM contact : (email first, Antispam default deny!) icq://27-83-87-867 tiny homepage : http://fromani.exit1.org (see IDEAS if you want send code!) known bugs : http://tcfoundry.hostme.it/mantis (EXPERIMENTAL)
