MPEG-4 to MPEG-2 using CCE 2.50 OPV: How not to waste bitrate?

[Fran-K]

I'm going to make a set of DVD-Video from a collection of animes in OGM format (video is DIVX 5). The video quality is good (200Mb for a 22 min clip). I always encode using the OPV option in CCE 2.50. I don't want to introduce more artifacts but don't want to waste filespace either. I'm realy bad for visual analisys of the encoding. I made 4 encoding tests with Q=20, 30, 40 and 50 but can't see any difference.
Is there a relationship between the 2 encoding methods?If a DivX clip with bitrate = 1246 and a Frame quality = 0,169, how much should be the bitrate / FQ of the MPEG-2 encode?

[name123]

I had that problem to, there is a bitrate calculator out there, just google it, its an excel file where it has the formula to calculate the desired bitrate to fully use the space on the disk. it use the # of frames, fps, and desired size and it will pretty much give you an accurate bitrate to use the disk space, once you find the calculator and get you desired bitrate i suggest you use that bitrate and do a 3-pass vbr with linear, zig zag, auto dc ticked, and image priority set to 5 or 10

[manono]

Hi-

I made 4 encoding tests with Q=20, 30, 40 and 50 but can't see any difference.
Is there a relationship between the 2 encoding methods?

Which 2 encoding methods? I saw you mention only 1, CQ encoding. Of course, the lower the Q the more faithful to the source the encode will be. Since you're beginning with a severely degraded source (your MPEG-4 DivX AVI), though, that may explain why you can't see any differences between the encodes.

i suggest you use that bitrate and do a 3-pass vbr with linear, zig zag, auto dc ticked, and image priority set to 5 or 10

I hope you meant to say "non-linear" quantization scale type, since virtually every decent retail DVD ever released uses non-linear. The very good HCEnc doesn't even allow you a choice, but always uses non-linear. And I wouldn't trust "auto" anything. For best quality tick 10 bits Intrablock DC Precision, if that's the setting in ver 2.50 to which you're referring. I don't have it, so I don't know exactly what it says.

[Fran-K]

Folks
I'm refering about MPEG-4 x MPEG-2. In other words, if a clip compressed in DivX have X bytes, how much bytes should have the same clip (re)compressed to MPEG-2 for the same quality?

I think that in a uncompressed clip, it should be about twice the size, but on a DivX clip maybe 30-50% is sufficient.

Thanks for the tips on CCE, i'll play with that.

[Midzuki]

Fran-K wrote:

In other words, if a clip compressed in DivX have X bytes,
how much bytes should have the same clip (re)compressed to MPEG-2
for the same quality?

It depends.

[jagabo]

I made 4 encoding tests with Q=20, 30, 40 and 50 but can't see any difference.

Look at enlarged still frames. The main difference will be the appearance of more macroblocks at the higher Q settings. 4x enlarged (nearest neighbor) examples:





These are enlarged crops from a moderate quality 640x352 Xvid AVI enlarged to 720x480 MPEG 2 with CCE. So some of the artifacts comes from the source.