INTERESING CCE 2.5 CQ ISSUE (possible bug?)?

[MITSUI_1]

I have been working on some VHS conversions, which I'm converting to SVCD using CCE 2.5.

So, I encoded the video using one pass VBR (constant Q), and set the Q to 26, the minimum bitrate to 768 and the maximum bitrate to 4096.

Here's the thing. The video turned out great, just barely fitting an hour of full screen pan and scan video per CD (I did use borders to save bitrate).

Anyway, here's the thing: I loaded up the resultant MPG in bitrate viewer and it shows peaks of up to 4937kbs!!! That's a lot more than the "max" bitrate I set, and just about the max of what my Apex could ever possibly handle.

Why did CCE disregard my max bitrate limit. Is this a bug in version 2.5?

[Truman]

This is nor really a bug. All MPEG encoders are more or less inaccurate. One pass encoding is not cce's strong point. I believe you will get lower peaks if you do a multipass encode.

[MITSUI_1]

Yeah, but the problem is that I get lower quality in general in the bitrate ranges that would be necessary.

I'm doing SVCD at 352x480, 24 fps NTSC film, with 32 pixel black borders on the top/bottom and 16 pixel black borders on the bottom, which helps the encoder immensely in terms of budgeting bitrate.

Even so, when I use multipass with an average bitrate of 1674 (and audio at 128kbs), as necessary for 1 hour of footage per CD, I see a lot of compression artifacts (i.e., fuzz around initial film credits, or near sharp lines). My sense is that no matter what you do, this bitrage -- if the movie hovers there most of the time, just isn't enough to encode the screen real estate of a "full screen" source (it works fine with letterboxed titles, provided that the image is resized to have the edge of the letterbox allign perfectly with a macroblock border).

When I use one pass with a constant Q factor, and lower it down to say 30 or 35 (remember that lowering "Q" in CCE gives BETTER quality), I can usually fit an hour of film on a CD and the quality is GREAT!

Somehow, even with peaks going all the way up to 4937, I got an hour of VERY nice footage on 1 CD. Interestingly, bitrate viewer reports that my average bitrate is actually much higher than 1674 (which seems mathematically impossible for a 1 hour picture, but maybe not). No macroblocks whatsoever, even during some very high motion scenes (we're talking Raiders of the Lost Ark here).

I guess I'll have to experiment with a few different settings and see how the software responds.

[mikehg]

What bitrate viewer are you using? I noticed that when I encode SVCD's at 23.976 with pullup enabled; that Bitrateview by Teco seems to make it's calculations on 29.98 which does not come out right. See what time length your bitrate viewer reports?

Take care,
Mike H.

[MITSUI_1]

AHA!

That might be right! I did run these clips through bitrateviewer AFTER I had run pulldown.exe on them.

That seems like it could be the reason for the change.

[Rav]

I am also facing the same problem with CCE 2.5. I encoded a mpeg2 (multipass) with a max rate of 2500 (min 500, avg 2100) but bitrate viewer shows peaks of more than 3000k. My player can not play SVCD (Sony DVP-S360) so I use TMPGEnc trick of putting VCD headers on mpeg2 file. Max bitrate that my player can play XVCD is 2750. When I play the movie, its fine mostly except a jerky play at few points. I guess its because of the over limit bitrate peaks.

I need to know is there a way to avoid these over limit bitrate peaks. I just started using CCE and is new to it. I used DVD2SVCD for encoding this file.

[Kdiddy]

Yes I wouldnt use bitrate viewer as the overall judge, Ive seen some inaccurrate readings from it.

Secondly, you should never get worse quality with multipass. You can adjust the Q levels in the advanced mode with this option as well.