Speeding up DVD2SVCD's QC test (TEMPGenc)

[Caffeine addict]

Hi all,
I am wondering if there is a way to speed up the QC test rate that DVD2SVCD sets -- without really messing everything up. Dvd2Svcd starts out at: Offset QC Value 80 and then TEMPGenc goes through 3 to 4 test runs (about 14min each). My question is: Is there a way to speed up the test runs or could I use a fixed QC rate instead? If I can use a fixed QC rate, how can I find out what they are for different DVDs?

Dvd2Svcd ver. 1.1.3 build 2

Thanks in advance!

[r6d2]

Originally posted by Caffeine addict
I am wondering if there is a way to speed up the QC test rate that DVD2SVCD sets -- without really messing everything up. Dvd2Svcd starts out at: Offset QC Value 80 and then TEMPGenc goes through 3 to 4 test runs (about 14min each). My question is: Is there a way to speed up the test runs or could I use a fixed QC rate instead?

Well, yes... and no. Two parameters on the INI file define how much testing DVD2SVCD will do:

Frames pr spread=25
Spread=100

I've not read anywhere what they mean, but my guess is this: The first one defines how many frames each clip on the sample will have, and the second is how many clips the sample will have.

All prediction-phase programs work with a (hopefully) representative sample of the movie in order to guess a CQ (TMPGEnc)/Q.factor(CCE) value for the whole movie.

You can change these parameters on the INI file to other values and shorten the prediction phase, but it is risky. Your sample may become non-representative.

DVD2SVCD is quite accurate with the default values and mostly underestimates (which means the last CD won't be filled, but not overflown also).

If I can use a fixed QC rate, how can I find out what they are for different DVDs?

If you fix the CQ then D2S will not predict. You will end up with an encode whose size is not foreseeable (not good to fill CDs to the edge).

Anyway, when going for CQ you will get a very good and stable quality on the encode, with less time than 2 pass, but the price you pay is prediction phase time.

You may also start with a different CQ and D2S may converge faster, but 80 is a good value anyway. CQ should be between 70 and 90 for a good encode (IMO). There is no such a thing as a Movie->CQ table you can look up. CQ depends (heavily) on the source, on the resize, on the filters, etc.

I think that's the way it is, unfortunately.

Hope this helps.

PS: Too much caffeine may be the cause for you to expect a big speedup on non-linear encoding tools! :D

[Caffeine addict]

Thanks a million, you have just confirmed what I had suspected (trust dvd2svcd :D ) and quite knowlegeablely too!

And your ending note had me cracking up! Good humor is alway a little truth based.

Thanks again!

[canoner]

Originally posted by Caffeine addict
Hi all,
I am wondering if there is a way to speed up the QC test rate that DVD2SVCD sets -- without really messing everything up. Dvd2Svcd starts out at: Offset QC Value 80 and then TEMPGenc goes through 3 to 4 test runs (about 14min each). My question is: Is there a way to speed up the test runs or could I use a fixed QC rate instead? If I can use a fixed QC rate, how can I find out what they are for different DVDs?

Dvd2Svcd ver. 1.1.3 build 2

Thanks in advance!

14 min each test seems extremly long. Usually each test takes out 1:40 to 2 minute clips and finish in less than 3 min when I choose "high" for prediction. Choosing "normal" will be faster and predict the size just as well. There is no visible gain with going "highest", as least I can't see it.