Monday, July 26, 2004

Silent Mobius

So far I've seen the first 14 episodes (out of 26) of Silent Mobius. The first episode is good, the second one is so-so and then it's a rough run up until episodes 12 & 13. Some of the early episodes are the formulaic "character introduction" episodes where we're given the characters background, usually while they're fighting a monster by themselves. We also have the required episode where the newly discovered heroine who's unsure whether she wants to be part of the team or not.

However, things do finally pickup the pace in episode 12 (Sister) when we start to learn about the background of the Lucifer Hawks and that certain characters are not what they seem. Episode 13 (Category IV) was good as well as things begin to fall apart and the battles between the two sides become more intense.

So if you're like me, stuck in the doldrums about eps 3-11, go ahead and stick it out because it does start to get better in 12 & 13. Hopefully the 2nd half of the series keeps up the pace and proves to be worth watching.

Update: Episodes 15-20 are pretty good. Things have definitely taken a darker turn in the series as it looks like the day of judgetment is now imminent. Episode 16 (Labyrinth) was actually a well done story focusing on Saiko Yuki. Episode 18 (Domull), Episode 19 (Back of a Coin) and Episode 20 (Love) are dark and troubling as the Lucifer Hawks really turn up the heat on Liquer Katsumi. The ending of episode 20 is a surprise and a nice plot twist.

Update #2: Hmm, plot holes. What happened to the sword that used to call Katsumi, "My Master"? We've seen the old dagger that Katsumi started with and Lum Cheng still has her sword. Katsumi and the named sword used to be inseperable in the early episodes, yet it wasn't in the apartment with Roy and Katsumi in ep 20.


posted by Wuphon's Reach at 8:36 PM (0 comments)

Sunday, July 25, 2004

TMPGEnc Express 3.0

Upgraded to the new version of TMPGEnc (an MPEG2 encoder) this week. The newer interface is nice. They've broken the batch encoding tool out to a 2nd application, which can run in the background while you prep additional files to be encoded. They've also fixed some problems with the old UI and you can now encode audio straight to AC3 (which makes the disk estimates accurate).

I'm still doing some benchmarking, but speed-wise it seems to be around the same. On my AthlonXP 1800+ box (PC2100 memory), encoding with "High" motion search precision and DAC of 10 bits, a 720x480 clip takes the following timings:

143 minutes in 16.5 hours (PCM audio only) or around 6.9 minutes to encode 1 minute of video. That's pretty close to the old v2.5 (which I used to estimate at 7 minutes to encode 1 minute). I didn't have the AC3 plug-in installed when I did that test which is why the output audio was PCM. Encoding to AC3 will probably add 30 minutes or so (not much).

A second session (this time encoding to AC3 audio) managed to encode 145 minutes in 17.6 hours of CPU time (or around 7.3 minutes per minute of source).

A third session (720x480 at 4000VBR + 224AC3) encoded 115 minutes in 13.5 hours (around 7.0 minutes per minute of source).


posted by Wuphon's Reach at 1:46 PM (0 comments)