Where is the line between choosing MPEG1 vs MPEG2?

[DRP]

Silly question maybe I know, but I usually only make SVCDs spread across two discs. Recently though I have needed to squeeze entire movies onto one disc for a 5 y.o. niece (she doesn't take having to change discs half way through very well).

This has forced me to start using VCD instead of SVCD. I always try to use the complete capacity of the disc, so I'm making XVCD wherever possible. This has meant making some movies in MPEG-1 with a 1350 - 1400kbps bitrate just to use up the available space on the disc.

Where is the bitrate point at which it is worth changing from MPEG-1 to MPEG-2?

I suppose that MPEG-1 is tuned specifically for an 1150kbps bitrate seeing as how that is the standard VCD and I've read that MPEG-2 quality is questionable below 1600kbps. What about the gap between 1150 & 1600 then?

[ZippyP.]

It's not really a choice between mpeg1 and 2 so much as a choice of bitrate vs. resolution. To clarify, mpeg1 and 2 at the same bitrate and resolution will look the same.

Resolution:
VCD is 352x240 and SCVD is 480x480, so SVCD has 3X the number of pixels. There is also CVD which is 352x480, 2x the resolution of VCD. CVD is created and burned the same as an SVCD, just with a lower resolution.

If you use TMPGEnc then you can get CVD templates for TMPGEnc in the Tools section of this site. D/L and put them in the TMPGEnc templates subfolder in order to use them. They'll appear in TMPGEnc next time you open it.

Bitrate:
The bitrate needed to keep a video looking good is directly proportional to the resolution. It's also sometimes called the bits per pixel ratio. What your tolerance is for bitrate limits is something you need to decide, as everyone has a different idea of what looks good. As well, using VBR encoding can preserve quality while allowing more minutes per disk. It will allocate more bits to high motion scenes that need it most, while saving bits from low motion scenes. I use 2-pass VBR in TMPGEnc when disk space is a problem.

Using 2-pass VBR here is my lower limits for average bitrate:
VCD 800 Kbps
CVD 1400 Kbps
SVCD 1700 Kbps

The minimum should be set as close to zero as possible, I use 250 because of my DVD player. For the maximum bitrate ( and for best compatibility) use the VCD or SVCD max. rates, 1150 for VCD or 2550 for SVCD.

Hope that helps a bit.

[ZippyP.]

What about the gap between 1150 & 1600 then?

I think I answered that but.....CVD!

[xtreemkareem]

IF it's a five year old, they won't know the resolution difference between MPEG-1 320X240 as opposed to MPEG-2 352/480X480. HOwever, insufficient allocation of data WILL be noticed if it results in macroblocking. Trust me. A xVCD at 1400 kbps though less than half the resolution of a SVCD will look 10 X better than an SVCD at 1400 kbps. Frankly. VCD will look rather decent on TV if you allocate enough data and takes less room than SVCD for equivalent quality. Go VCD MPEG-1 352X240 at higher bitrates if your player hadles it. On PC, VCD will even look alright as long as the bitrate is high enough. It also helps to sit 10 feet back from the monitor, but what the hey.

[DRP]

Okay, thanks everyone. The video I was converting was 96min long and needed to be on one disc. I looked at the KVCD option but it's really too difficult bitrate calculation wise. I need much more filesize predictability than appears to be possible with the KVCD method, so I chose to stick with slightly more standard formats instead.

I ended up converting the video to PAL (to shorten the running length a bit) and then sacrificing a lot of the audio bits and using them instead for the video.

The end result was an MPEG-1 XVCD with an average VBR (0-2500) of 1147kbps and an audio bitrate of 160kbps. I then overburnt that to a 90min CD-R. Thankfully it plays fine in the DVD player. I'm sure the end result will be good enough for the 5 y.o.

I had forgotten about CVD. I'll try that the next time I need to do something that works out as being between 1150 & 1600kbps.

Cheers

[xtreemkareem]

Dude, I too love CVD, but it's an insult to the video IMHO to go below 1700 kbps w/ CVD. That's the equivalent of 850 kbps VCD MPEG-1. Trust me, keeping the CVD bitrate in the upper limits (2000 kbps +) is best for decent quality CVD.

[DRP]

Perhaps someone could clear something up for me?

ZippoP said there's no difference quality wise between MPEG-1 & MPEG-2 at the same bitrate and resolution.

If that's true, then why was MPEG-2 developed? What's the point? In fact, why would anyone bother buying TMPGEnc Plus, when it would be perfectly possible to make non-standard SVCDs (or VCDs) with SVCD resolutions but with MPEG-1 compression using the free version of TMPGEnc?

I thought MPEG-2 was a younger and more developed compression algorithm and therefore better in quality. So if you compared equivalent bitrate:resolution samples, then the MPEG-2 one would be better quality.

What am I missing here?

[ZippyP.]

Mpeg2 has more features, read here.

...it would be perfectly possible to make non-standard SVCDs (or VCDs) with SVCD resolutions but with MPEG-1 compression using the free version of TMPGEnc?

It's possible, but not many DVD players will accept it.

[DRP]

Dude, I too love CVD, but it's an insult to the video IMHO to go below 1700 kbps w/ CVD. That's the equivalent of 850 kbps VCD MPEG-1.

I'm not sure this can right. MPEG-2 must be more than simply a compression algorithm for higher bitrates than the standard VCD? There must have been some tweaking and efficiency gains between the development of MPEG-1 & MPEG-2?

You are saying that the resolution difference between CVD & VCD is 2:1. I understand that, but then you are using the same ratio to come up with the equivalent bitrate. Is it fair to do that? I would have thought there would be some efficiency gains in MPEG-2 over MPEG-1 which would skew that. For instance, standard SVCD at 2500kbps does very definitely look vastly superior to standard VCD. The ratio difference in pixels is 2.73:1, yet the same in bitrate is only 2.17:1. There must be something in the way the compression algorthim works to make up that difference (plus some more) in order to overcome that deficit in the bitrateixel ratio.