So, according to Thoggen’s count, it took 4.45 hours to encode into Theora. At least with this DVD it seems that the coding team has fixed their error where the software seems to keep going past the end of the DVD. As far as I recall from one year ago, it didn’t add more infomation in – the thread just didn’t get the message that the title was done encoding.
I have not watched the file yet as I went straight into encoding the DVD into xvid. According to ls, the file took 706600062 bytes or 674 Megabytes. I told it to target 650 – but these algorithms are notorious for not being able to get it just right. I think it’s a general programming problem – not something that these guys just haven’t figured out because they are lazy. So, I would say, if you have a 700 MB CD-R (pretty much the standard nowadays), tell it to aim for 650 and you’ll be ok.
Now, for the Xvid side of things, I wanted to be fair to the Thoggen folks. Just as I didn’t use libtheora and some command line tools to encode the file (which may have been faster or different in some way), I decided not to encode the Xvid file with raw mencoder on the command line. I used dvd::rip, a perl frontend to transcode. It is pretty complex and capable of encoding to avi, mpeg, and ogg containers. I told it to try and reach a target size of 650. I also told it to rip subtitles (since it can) and so it predicts that the filesize will be 651 MB. We’ll see how close it ends up being.
Specifically, I told it to use Xvid4 with MP3 128kbit/s audio. I’m not sure how that compares to the Vorbis encoding Thoggen chose. Again, Thoggen is very Gnome-like and doesn’t provide too much configuration. It’s both a blessing and a curse. I told it to use smart deinterlacing as I didn’t know if the video was interlaced or not. I only gave the documentation a quick read today, so I’m not sure if there’s a way to encode the file in only one pass. That’s usually what I do with mencoder on the command line. Apparently the two pass encoding seems to make it take just as long as Thoggen. It appears to be taking somewhere in the vicinity of 4 hours to encode. Again, with my command line settings it usually only takes 20-30 minutes – but I wanted to compare automated solution to automated solution. (And, with ffmepg2theora, I think it also takes somewhere in the realm of 20 minutes)
So, I think the file may not be ready for me before I hit the hay so join me tomorrow for our conclusion in part 3.