PDA

View Full Version : Bitrate vs. Motion Search Estimate


mguru
26th April 2002, 01:37
I have an older system (PII-400/128mb) and it takes a 120 min movie about 133 hours to encode with 400/2300/2500 2xPass bitrate with Highest Quality. With same bitrate setting and on High Quality it cuts that in half to about 64 hours. Automatic CQ on Highest Quality with 2300 is 91 hours and on High Quality, 58 hours.

First, is there a really big difference in High and Highest quality setting? I have done tests with both and the Highest is a little quieter with pixellization but it is almost unnoticeable to my eyes except in fast moving scenes or in loud backgrounds. My preference is quality over time. I don't mind waiting for 4-5 days for it as long as its almost lossless from original. But if I can get a nearly identical outcome in half the time, then of course that would be the best........What is everyone's opinion on this?

Second, how does bitrate and motion search estimate both affect end quality? For instance, if I just use High quality but up the bitrate average 100-200bps (for 2x or Auto_CQ) would I get the same result as if I had had a lower bitrate and Highest Quality?

Third, I have noticed that even with the new 2xPass, I still seem to get overall better quality with Automatic CQ. Anyone else see this as well? However, I am averaging a 10-40% decrease in the end file size with VBR, which I guess is the whole point of doing it that way. No need to debate this...I guess its a purely personal preference and has been debated here and elsewhere a million times over.

Thanks,
mGuru

WarpEnterprises
30th April 2002, 08:28
Hi!
The main advantage of 2pass is that you can precisely set the average bitrate = the resulting file size, so you can aim at exactly filling one CD with your video.
You have to "trust" TMPG for deciding to use the optimal encoding regardless if CQ, VBR or 2pass-VBR - the encoder has some degrees of freedom where exactly to spend the bits.
I meyself noticed no real difference between high and highest quality, the Q-factor seen (with the programm bitrate viewer) also differs only very slightly (but sadly this Q isn't the whole truth!)

mguru
30th April 2002, 18:10
Thanks for responding. I'll keep screwing with it. Until I can upgrade, I think I'll just go the HQ route since it seems the difference is negligible and 2x as fast.

BTW, if anyone hasn't noticed, a new TMPGEnc version is out: 2.54.36.134

mGuru