MPEG-2=better compression, right?

[e-z-e]

i read on this site that mpeg-2 encodes have better compression, hence smaller file sizes, than mpeg-1. i was wondering how. i was guessing that the longer encoding time allows better compression to take place, but im not sure.

[Vejita-sama]

Ok, this is kind of long and more info than what you asked but I'm making some assumtions as to 'why' you asked the question so:

No, in fact MPEG1 and MPEG2 have yield (virtually) the same compression. The size of your encoded MPEG (be it MPEG1 or MPEG2) is 100% dependant on the bitrate and the source run time.

At the same bitrate and resolution MPEG1 looks the same as MPEG2 (video quaility wise). The reason that MPEG2 is considered superior to MPEG1 isn't because of better compression or video quaility, but because it supports more features: interlaced source, selectable subtitles, multiple audio tracks, etc. etc.

The VCD standard calls for MPEG1 352x240 video=1150kbit/s audio=224kbit/s
The SVCD standard calls for MPEG2 480x480 video=2520kbit/s audio-224kbit/s.

But if you made an xVCD at MPEG1 480x480 video=2520kbit/s audio=224kbit/s; quaility wise, it'll look the same as an SVCD. But you can encode MPEG1 much faster than MPEG2. Plus more players (and all computers) support MPEG1, so a lot of people prefer to make xVCDs over SVCDs.

xVCD or xSVCD doesn't mean better, just non-standard. Quaility can be better or worst depending on the settings you us. Basically the higher the bitrate and resolution, the better the quality of the encode.

Now as to smaller file sizes. You have two methods of encoding:

CBR - constant bitrate, the same bitrate is used for the whole encode
VBR - variable bitrate, the bitrate raises and lowes during the encode. Basically the software lowers the bitrate for low motion scences and raises it for high motion scences.

Older verisons of TMPGenc have VCD set for CBR, and SVCD set for CQ_VBR (constant quaility VBR) [newer verison of TMPGenc tend to have SVCD set to CBR too). Anyway, if you use VBR it's possible to enocde longer video clips and still have good quaility, at smaller sizes.

eg. I have a 90min runtime source. But I want to put it on 1 80min CDR. In order to do this I use a bitrate calculator and find that I can use video=1051 audio=129kbit/s. If I encode CBR at video=1051 some high motion scenes will look like crap.
But I could encode VBR, with min=300, max=2520, ave=1051. This will lower the bitrate to as low as 300kbit/s, raise it to as high as 2520kbit/s, but do so such that the aveage=1051kbit/s. That is I'll get the same size MPEG at the end. But because the bitrate will increase in high motion scences they'll still look nice (it balances the increase by decreasing in low motion scences).

You've got 2 VBR choices in TMPGenc: CQ_VBR and 2pass VBR. CQ_VBR works as discribed above. 2pass VBR first makes a CBR pass, then goes back and uses the CBR encode to make a VBR pass, this allows better allocation of bitrate and better control of the final MPEG (truth is no one really knows how CQ_VBR works ). But as the name/methods states it makes 2 passes, so it doubles the encode time.

[e-z-e]

i appreciate the detailed reply. i have one more question though. my dvd player sometimes has trouble playing non-compliant vcds/svcds so will altering the CBR/VBR change the compliancy of my svcd? i know if i opened tmpg and loaded the svcd template, all the settings i can change would not alter the compliancy. but i dont have tmpg on this computer, so i dont know if it is one of those changeble settings.

[kinneera]

For SVCD, both CBR and VBR are compliant, as long as the video bitrate remains below 2520. For VCD, only CBR at 1150 is compliant. CBR at any other bitrate is XVCD, as is VBR.

You don't have to confine yourself to the changeable settings in TMPGEnc. Just load the unlock.mcf template to gain access to all the options.

[hpv]

Actually, MPEG-1 and MPEG-2 at identical bitrates will not have the same quality. You might think that someone standing there saying so in a long-winded explaination does so because they know what they're talking about, but in fact they are terribly misinformed.

[mr_seth_enslow]

Care to elaborate then?

[Nelson37]

This has come up before, and while I believe most of the previous comments are correct, the "equal quality at equal bitrate" I must disagree with. On 2 seperate clips at IDENCTICAL bitrates below 1500 (only change was mpeg-1 or 2), my Mark I Eyeball detected superior quality on MPEG-1 with nearly identical filesize.
It is my understanding that MPEG1 is optimized for low bitrates and MPEG-2 for higher bitrates, while this may not be correct, the eyeball test seems to confirm this.

[kinneera]

This is more a case of certain encoders being better at one type than another. The fundamental underlying mechanisms for eliminating data during compression are functionally equivalent. You also must match all parameters, especially resolution, not just bitrate.

[Vejita-sama]

At the same bitrate and resolution MPEG1 and MPEG2 are the same.

HPY - please tell me how this statement is incorrect. Have you read the white book standards for vcd and svcd, and look at how the compression works?

Nelson37 - were the resolutions the same? and both set for CBR w/ the same encoder. I'm not saying your MKI eye ball is off. Also remember that we're doing software encoding, not hardware encoding so there could be some slight differences introduced.