CQ beats VBR bigtime...

[randallc]

Adam, if you're speaking of VBR equaling the quality of CQ within a constrained bitrate limit then I agree.

But that's not how CQ is best used as I said previously. CQ is idea when you don't have a (low) limit on the bitrate, then it truly shines. I don't care if you have 10 pass VBR, if the vbr is limited to 2600 bitrate there is no way in the world it will come close to matching that scene when the CQ is using 9,800 bitrate.

I know, you're thinking well that's not fair to compare them at different bitrates, but that's how CQ is intended to run, you select the level you want and it uses the bitrate that it needs to maintain that level. This is why CQ is not an ideal choice for encoding DVD's into SVCD's - CQ doesn't operate near its potential with such a small bitrate upper limit that the SVCD specs impose.

If you're encoding for DVD definitely use CQ but for SVCD and you're looking for the best quality in a small file, multipass VBR is the way to go.

Randy

<font size=-1>[ This Message was edited by: randallc on 2001-11-07 10:49:51 ]</font>

[adam]

Well if you were encoding a dvd than multipass vbr wouldnt be designed to have a max bitrate of 2.6Mbits either. Like I said at comparable settings CQ and multipass vbr are identical in quality. With the exception of speeding up encoding time, there is absolutely nothing that CQ can do that multipass VBR can't.

CQ is used to speed up the encoding process when compression isnt a concern, period. Actually CQ is much better for svcd than dvd because since your max and min provide such a huge interval with dvd you see the benefits of better bitrate allocation much more.

[randallc]

Not if you select a fairly high Q level. It the video will get choppy. CQ will be wanting a high bitrate but because its limited to svcd limits the video will end up looking bad.

This is a case where VBR outshines CQ for SVCD.

[SatStorm]

Let me also post something here...
I feed tmpeg always with mpeg 2 files, after proccessing them with dvd2avi. Those mpeg 2 files are rips from satellite transmissions, exactly the original stream.
I used 2 Pass VBR and CQ_VBR. By Using CVD (352X576) I get better results on same bitrates.
I have to say that CQ_VBR gives me better quality on some version of TMPEG. On others, 2 Pass VBR is better.
Overall, 12H is better for CQ VBR, while the new one (2.1) is better for 2 Pass VBR.
Old 12 created smaller files in both settings, while 12a was working only with CQ_VBR (2pass was buggy).
What is my point: Seems like the source and the encoder makes some difference on some things. Also the settings: Field order, interlace, original resolution.
In theory, 2 Pass VBR should alway be better, because that fuction also includes "detect scene".
But it seems that TMPEG authors was use to focusing on CQ_VBR, maybe because of speed. So, by using CQ_VBR, Detect scene change and soften block noise together, you get the same results as 2 PASS VBR on most cases with earlier tmpeg versions, but in Half time... That change with new TMPEG versions, which speed improved (but the quality of CQ_VBR get worse)
Anyway, I use CQ_VBR mostly, so I stuck with 12H. But when I have a difficult movie, I give a try to 2Pass VBR. It takes me a day more, but I can choose the better result to keep.... 2 Pass VBR with 2.1 seems very promising
Let's hope in a near future version of TMPEG, CQ_VBR quality will improve. Even if the speed improved, I prefer not to wait twice for 2Pass VBR, expecially if the quality is so near...