Can't get proper quality?

[RoopeT]

Ok, I have made a few SVCDs with CCE before, but with this DVD I just can't get the quality I want:

DVD: AC/DC - Stiff Upper Lip - 16:9 2 h 7 min PAL interlaced

Tried with these settings: fit to 3 x 80 min CDs

Bilinear resize to 480 x 576, Audio 224 kbps, video 2300 - 2490 kbps, one pass VBR Q 1, Image quality priority 20, no noise filtering (I think the source quality is really good). Linear quantitizer, progressive, Zigzag, upper field all unchecked.
(no matrix patches, CCE 2.5 SP)

The result is really close (within a couple of M) to 3 x 800 cd images, BUT:

There are still some macroblocks with these settings, and I really cannot figure out what to do to get better quality! Is there just so much motion that I have to resize to 352x576 to get decent quality within the SVCD bitrate limits?

[chainsaw135]

Have you tried 4 pass vbr setting?

[RoopeT]

Haven't tried 4 pass, since it won't make any difference when I'm already using nearly the maximum bitrate. Using 4 pass (max 2490, min whatever) won't give you better quality than CBR @ 2490, just a smaller filesize. However, I'm not trying to fit this on 2 CDs, and with 4 CDs I would have a bitrate of about 3200, which I'm pretty sure my player won't handle.

[Mozart]

he he... it was my first assumption too. I had think "why to use multipass vbr if I am using a very high bitrate?". Well... there are movies and movies...

examples:
a) "the wall". Method: 1passVBR, min=300, max=2400, Q=1. Result: very smooth picture. Realy good;

b) "cast away". Same method of a. Result: 95% of the movie was good. 5% has a lot of bloks. When using 3 pass VBR, min=300, max=2400, avg=2000, the result was awesome (and the file size smaller, of course).

conclusion: if you need a 100% good encoding with CCE, it is mandatory to use >=3 multipass VBR, and doesn't matter the max bitrate used.

[RoopeT]

But there's something different with our one pass settings: I used a min bitrate of 2300, not 300.

Actually, if I used min 300 setting, there would be a risk of creating an mpg that doesn't play on my Philips DVD. I have notised some skipping when the bitrate goes below 800-1000 kbs. When using standard 480x576 resolution, this is very unlikely to happen, but happened for example when I made a 1 CD XSVD (352x576) copy of the Tigger Movie (74 mins, 4 pass avg 1350, max 2000).

[chainsaw135]

Yeah I was just figuring since you were looking for something to try that "might" help maybe you should not over look the muliti vbr methods. Maybe after you've tried that method with this particular movie you could let us know if it has helped you out.

[Mozart]

Originally posted by RoopeT
But there's something different with our one pass settings: I used a min bitrate of 2300, not 300.


ok, this is another thing that you need to keep in mind while making a SVCD: when using 1passVBR with Q smaller than 5, CCE behaviour is the same as if it was using CBR=max bitrate. What I mean: lets suppose 3 different encodings of the same movie:
a)1passVBR, min=300, max=2400, Q=1;
b)1passVBR, min=2200, max=2400, Q=1;
c)CBR=2400.
If you make the test, you will see that
file_size_of_(a)=file_size_of_(b)=file_size_of_(c).

Thus, for a low Q value, while encoding SVCD, it doesn´t matter the value of min. It will not use such value. This probably is not true if making DVD, because it will allow a 4x highest max value.

[chainsaw135]

Mozart,

Thanks for the information something I didnt know myself, will keep a mental note of that.

[NearlyCritical]

Originally posted by RoopeT
Haven't tried 4 pass, since it won't make any difference when I'm already using nearly the maximum bitrate. Using 4 pass (max 2490, min whatever) won't give you better quality than CBR @ 2490, just a smaller filesize.

Finally, someone else realises this :)

Why waste all those hours doing a multiple pass encode when it makes no difference to quality? (assuming file size isn't a concern).