http://renomath.org/ejolson/video/avchd/
The creation of the high definition mpeg2 files from AVCHD source runs at about 15 frames per second on my computer. This means transcoding AVCHD to a format Cinelerra can edit is slower than capturing HDV from tape. AVCHD source can be losslessly authored to BD5 with reencoding necessary only for rendered effects and transitions. Therefore a workflow with essentially zero transcoding is possible. See Lossless AVCHD Editing in Linux for more information about such a workflow.


So what about that lossless editing?

http://renomath.org/...chd/lossless.html
Do not use any transitions or filters. For transitions and filters refer to Editing AVCHD with Cinelerra. [AKA: recursion: see recursion]


This might be changing though:
"I've been working like hell on patching Cinelerra," he told me, as well as using other tools, such as Blender. Montgomery sees one of his short-term goals as getting the overall quality of FOSS video tools improved, while he tackles his more forward-reaching goal: opening up the untamed frontier of video development.

http://www.itworld.c...r-despite-mpeg-la

This is a great development, but the timeline doesn't help me any. Maybe it will pay off in time for the next upgrade.