Single pass vs 2 passes?

[chupa]

I currently using gordian knot to do divx 6.2 encoding.

Doing 2 passes takes a very long time and I'm wondering if there is really
that much difference in video quality between doing 1 or 2 passes.

keep in mind I'm backing up each dvd to 2 cds.
If I was just doing it to 1 CD I assume there would be a difference.

[redwudz]

That would depend on several factors. If you have a high action video, it may help with a little better encode. Of course, the length of the video and the encoder settings are very important. Why not try a short 10 minute representative clip from one of your DVDs and see how it looks with both settings?

Multi pass may be better, but the question would be, is there any visible improvement that is worth the extra time?

And welcome to our forums.

[chupa]

I would do that except odds are even if there was a significant quality difference I woudn't
notice it on my small monitor or small TV.

The quality difference would be easy to see on a big screen TV if there was any. And I
want these discs to look good on a 40 inch which I will get in the future.

[drewzor]

You can use a biterate calculator, I use aspect 2 but there is one in divx aswell. In aspect 2, I enter the pixel dimention, duration of video and audio bitrate and specify an output file size I'd prefer. It returns a bitrate for video and if I have set the file size too small it has a red dot to mark insufficant size. I then change the file size limit until it goes green meaning that it the right bite rate for quality

[manono]

Hi-

You can't do 1-pass encoding using GKnot as the front end. To do it, you'll need to do it yourself in VDubMod. But maybe you know that already.

Personally, I would never recommend using 1-pass CBR encoding for any reason. The static scenes will get too many bits unnecessarily, and the complex scenes will be bitrate starved and block up. If you care about quality at all, don't use it.

1-pass VBR encoding (Single Pass Target Quantiser, in XviD) is quite good, but you'll lose all control over file size. If you don't care much about the final size (keeping it on a media center HTPC, for example, or burning to DVDR), 1-pass VBR is probably the way to go. And it takes about half as long. But if you're burning to CD (need a fixed size), you'll have to stick with 2-pass encoding.

[GMaq]

1-pass VBR encoding (Single Pass Target Quantiser, in XviD) is quite good, but you'll lose all control over file size. If you don't care much about the final size (keeping it on a media center HTPC, for example, or burning to DVDR), 1-pass VBR is probably the way to go. And it takes about half as long. But if you're burning to CD (need a fixed size), you'll have to stick with 2-pass encoding.

Hi,
Although manono is quite right in stating that you lose all control over file size it can be controlled somewhat with frame resolution size, For instance if you encode a movie at 512x384 instead of 640x480 you may not have a visually noticeable quality loss but since the bitrate is spread over less pixels you will end up with smaller file in the end. This is another area where trying test clips is a good idea, If you are spreading 1 DVD over 2 CD's you have quite a bit of trial and error room to play with. Also the quantizer settings will have a profound effect on the end file size, For DivX and XVid I wouldn't recommend a quantizer setting higher than 3.

[chupa]

Well thanks for the help guys. I'm trying to keep my resolution close to 704x280 after
black bars are cut off, and bitrate at 1300kbps-1500kbps, so I think I'll just have to put up with the long wait of 2 pass.

[manono]

I hope you didn't literally mean 704x280. For a couple of reasons that's not a good resolution. One is that it gives you an AR of 2.51:1. There just aren't many movies with that AR. Even 2.35:1 movies will be too narrow, and people will look short and fat, and round things will be ovals. The other is that 280 is only Mod8 (divisible by 8), and you should be using Mod16 for compressibility reasons. MPEG-4 encodes by 16x16 macroblocks, and Mod16 resolutions give better compressibility. In GKnot the H-Module is set at 16 for a reason. Don't change it. Maybe for 2.35:1 movies, 720x304 is better. Let GKnot tell you, after you crop away the black and resize, keeping the AR Error low.

Also, you can't just set a bitrate of 1300 or 1500 or whatever and be assured of decent quality. Different movies compress differently, and since you're using a high resolution, all the more reason to run the compress test first. That's one of GKnot's greatest innovations. Or maybe you know that already, and are already using it.