On Wed, 2006-05-31 at 10:17 +0930, Jonathan Woithe wrote: > The biggest problem is that libdv's PAL encoder is buggy (I've never tried > NTSC) and introduces considerable artifacts which are particularly > noticeable during moderate horizontal movement. ffmpeg is better but still > (as of 6 months ago anyway) had problems. Besides this, there will be a > loss of quality due to the decode/encode cycle since DV is lossy. Most > videos do have at least some transisitions/overlays etc, so this issue will > crop up, even if it doesn't affect a majority of a video if the source is > DV. For this reason using DV as the intermediate format is not ideal. > > However, perhaps the DV framework could be used as the basis for something > since as has been pointed out, the DV export option is one of few (perhaps > the only one?) which does video and audio simultaneously. >
Perhaps the right solution to start is to develop a totally separate Render-Wizard application, which could do many tasks in a optimal way. It could be coded in something simple/rapid like python/perl/etc... and make use of batching through cinelerra. Thus, solving the immediate need to get SOMETHING, and allowing you to experiment and develop the optimal solution for final integration into the mother-applicaiton somewhere down the road. In this case, the challenge is to write an app that can read the XML and produce a "batch" config to achieve the desired outputs (both video and audio). I imagine it would generate a temp xml config to work off of. For NTSC DVD output, I still prefer rendering to raw RGB and then to mpeg2encode. For my snowboarding videos (lots of white snow), this yields the best/least distortion. Need lots of tmp space for redndered output (pre-mpeg2 encoding). Herman, would this be acceptable? If so, I might give this a go... -christoph _______________________________________________ Cinelerra mailing list [email protected] https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
