How much space needed for mpeg1, in comparism to mpeg2?
 

I've encoded the same movie of a lenghts of 42 minutes with the same settings:

CQ 100
min bitrate 300
max bitrate 2530

the same gop structure (kvcd template),
the same audio settings (48000 khz, 192kbit)

The first run I choosed mpeg 1 as output format (mpeg1 Video CD non-standard) and then mpeg2 (mpeg2 Super Video CD (VBR))

To my surprise the output size of the mpeg1 encoding is 740.198.648 bytes and the mpeg2 output ist 859.443.088.

So mpgeg 2 needs 15% more space for the same job?
Has anyone done similar experiences? I found nothing doing the search.

Do the same with CQ80 and you will see that MPEG2 does not need some much more space.

The CQ100 is very confusing there.

yeah. you're max cq should only be 90. after that there is no quality difference.

To understand your answers right I like to ask back:

Does the space difference decrease when I choose al lower cq-Level?

Or is it not right to compare the two formats with the same quality level (means e.g. mpeg2 produces with cq80 the same quality as mpgeg1 with cq100)?

You have to compare both method with the same CQ but with a CQ lesser than 90 (let say 80).

Because above 90, file size raise vey, very quicly and you can't really do any comparison with this. For instance, you found that the diff is 15% with CQ=100. But with CQ=99.99 the difference is perhaps 5 %.

With CQ below 90, you will have results that are more constants. And then you can consider the test as valid.

As far as I understand I still don't get the point of this CQ value thing (and I really tried to find a documentation with no success...)

I thougt the CQ value indcates how much quality loss I accept.
And further I thougt 100="no quality loss" and values below indicates the size of the compromise I'm willing to accept.

The matter seems to me far more complicated and mysterious than ever.

That would be true if the curve f(CQ,loss) was a line like y=x. But it is not. Above CQ=90, the curve becomes near flat. You increase the CQ but the quality is quite the same.

But the filesize of the mpeg becomes out of control ! :-)

You want to know how much more space MPEG2 need compared to MPEG1 or not ? If yes, do the test at CQ80. If not, just close the thread :-)

http://tangentsoft.net/video/mpeg/enc-modes.html

-kwag

Good link. I love this part :

O.K.
I've done the test as proposed. The same settings as described above, again 352*576 resolution. I took a clip with advertisment and trailer of 3:12 minutes, which has a in my opinion good combination of high and low movement parts.

Here are the file sizes:

CQ__________mpgeg1___________mpeg2________Index
100_______55.255.424________64.809.388________117
95________54.985.840________63.647.388________116
90________54.785.976________58.478.812________107
85________54.641.888________54.095.748_________99
80________49.803.320________49.715.008________100
75________45.283.140________45.062.360________100
70________38.908.408________40.284.216________104
65________37.844.016________37.509.360_________99
60________33.704.972________35.152.824________104
55________30.277.072________31.706.332________105
50________27.588.204________29.052.324________105


Hope I did no mistakes. Comments welcome!!!

In sum : between 0 and 5% more space needed for MPEG2 than MPEG1 when encoding with TMPGENC. That is more or less what I thought.

Strange to see how this depends on the CQ.

Between cq 65 and 85 (exept 70) mpeg2 file is smaller, why is that, hmm?

Thats normally the advantage of mpeg1 when encoding using average LOW bitrates like we do. 8) But I'm surprised of differenzes in your results at high/mis and low CQs 8O

Well I did also a testing and it gave me at almost all CQ between 40 and 80 the result that mpeg1 cames out at smaller sizes than mpeg2.
Maybe it depends upon the contents of the used movie? Hmmm ..