Chances are I've screwed something up completely but I've been trying to encode a 3-hour, high action film to fit on 2 CDs with the KVCD LBR template, and, so far, even when my test encodes are way above the predicted file size quality is not very good, with mosquitos around everything and blockiness whenever there's a high-action scene , even using Faerydust and Convolution3D. KVCDx3 has given better results, but not spectacular and the file size is way too big. Just to see what would happen, I encoded a test clip with 2-Pass VBR, and although it took much longer, it was much better quality and would fit easily on 2 CDs , with file size predition much easier being able to specify the average bitrate.

OK, so I think this just can't be right, as someone would have thought of using 2-Pass VBR before, and KVCD being much better than everything else size/quality wise...but I am currently encoding 4 test clips.

Clip1: 2-Pass VBR, Max=2000 Min=300 Average=1150
Clip2: CQ_VBR, Max=2000 Min=300 CQ=50
Clip3: 2-Pass VBR, Max=2000 Min=300 Average=1150
(Using New GOP and QMatrix)
Clip4: CQ_VBR, Max=2000 Min=300 CQ=50
(Using New GOP and QMatrix)


If anyone would like to tell me not to bother coz I've probably done something stupid and maybe suggest what I've done, or direct me to a thread where this kind of thing has already happened, please do!

PyRoMaNiA
in this link is all about CQ vs. CQ_VBR :

http://www.kvcd.net/forum/viewtopic....ca2df842301ff6

But I was talking about 2-pass VBR vs. CQ_VBR...

Anyway....finished that testing and the resulting files are not noticeably different in quality...although there is slightly more mosquito noise with the New GOP and QMatrix...

Clip1: 22.2 MBs
Clip2: 24.7 MBs
Clip3: 22.1 MBs
Clip4: 24.7 MBs (although it is actually 34 KBs smaller than Clip2)


By the way, the clip is 2:41 long.

So this means 2-Pass VBR is smaller than CQ_VBR?
Is it actually worse quality in anyone elses experience?

"But I was talking about 2-pass VBR vs. CQ_VBR..."
yes,of course my friend,
but we don't use 2-Pass VBR no more.
then i post a link....read the 4 final pages .

Hi PyRoMaNiA,

CQ wins every time against 2-pass. At least with TMPEG

-kwag

Oops . Figured out why CQ was lower quality...I was using the KVCD Predictor to get a sample size and had to encode at a low CQ setting...just used manual prediction and the predicted size was much bigger, meaning I can use a higher CQ setting...so the predictor was screwing up for some reason, well seems to be working now, thanks for your replies!