Log in

View Full Version : New Mod HQ


lighty
31st October 2002, 21:20
I read in new version of XviD guide that for using Mod HQ one have to set quantization type to MPEG in 1st pass and then to Mod HQ in 2nd pass.
However I've found while browsing that if one set Mod HQ in both 1st and 2nd pass then MPEG quantization type is automaticaly used in 1st pass and Mod HQ in 2nd.

So... what is truth?:confused:

If I set Mod HQ in both pass will it switch quant types automaticaly or not?

ffroms
31st October 2002, 21:27
As far as I know it uses MPEG in 1st pass and modulated in 2nd so u can use MOD HQ safely in both passes like I do.

Koepi
31st October 2002, 21:35
LIGHTY!

Read the forum rules. Rule 1 and [b] 1a [b] are there for a reason. You opened the Xth useless thread about it.

SEARCH! And read.

*gna*

:P

Koepi

lighty
31st October 2002, 23:28
@ffroms
Tnx for the info!

@Koepi
I've found some conflicting and inconclusive info in existing threads and the new XviD guide confused me more... I really needed to have that question unequivocally answered.

BTW- if the ffroms statement is correct then new guide unnecessarily add one step more.

Koepi
1st November 2002, 00:52
Hmpf.

Well, I answer this as usual in these threads, were the users are so ignorant and refuse to use the search and read up the threads which are dealing with the backgrounds:

If using modulated quantizer at 2nd pass, use:

Modulated:

- if the average quantizer in 2nd pass is going to be >3.0, use mpeg quant (used in 1st pass automatically since at quant 2 and 3 mpeg quant gets used). If the quantizer is e.g. 3.01 in second pass, then you get better results using h263 quant type in first pass.

- for new mod. hq it's exactly the same, just replace mpeg by h263.

Koepi

trbarry
1st November 2002, 05:29
Aha! I had been nodding my head and then ignoring all this sage advice now for months because I didn't understand it and was just using mod q on both passes with good results.

But I see now that I was really only getting away with it because I tend to waste bits to get high quality on my HDTV caps and the quants were probably < 3 anyway.

- Tom

Koepi
1st November 2002, 10:42
Hum, re-reading that post I think I made myself not very clear:

"Modulated" uses MPEG quant type for quants 2+3, thus if the 2nd pass quant will be >3 you should use h263 quant type in first pass.
"New Modulated HQ" uses h263 quant type for quants 2+3, thus if the 2nd pass quant will be >3 you should use MPEG quant type in first pass.

Regards
Koepi

lighty
1st November 2002, 12:56
@Koepi

Tnx for the explanation!:cool:

serbersan
3rd November 2002, 15:37
Originally posted by Koepi
Hum, re-reading that post I think I made myself not very clear:

"Modulated" uses MPEG quant type for quants 2+3, thus if the 2nd pass quant will be >3 you should use h263 quant type in first pass.
"New Modulated HQ" uses h263 quant type for quants 2+3, thus if the 2nd pass quant will be >3 you should use MPEG quant type in first pass.


@Koepi:

I don't understand very well why the limit to choose between MPEG and H263 is an average of 3.

I explain what I think about that:
For example, if you do first pass in a movie with
Mpeg--> 1400Mb
H263--> 1150Mb

Using h263 in first pass, If in 2nd pass you obtain an average like 2.5 you have used MPEG(using modulated) almost all the time. Hence you have used first pass h263 and the final curve in second pass it's more near to the curve obtained if in first pass you've used Mpeg, hence you have to use mpeg in firt pass to achieve better results.

But if this is correct, why to choose an average of quantizer 3 instead of percentage of quantizer type??? I mean:
If the percentage of the quantizer type used in first pass it's less than 50% in 2nd pass the resulting curve It's more near to a curve made with the other type of quantizer in first pass but the average could be more than 3. The last movie I made has an average of 3.3 and a 77% of mpeg using h263 in first pass and modulated not modulatedHQ.

And a last question: In any case, when you use modulated or modulatedHQ even if the quality is amazing the results or the scaled curve aren't as optimal as the results using the same type of quantizer in both passes, due to the different size between quant 2 mpeg and quant 2 h263??? It isn't the same like use a 2nd pass using a filter for example that you haven't used in the first pass???

I've searched over the forum but I haven't found an answer for those questions this is why I ask.


Thanks Koepi, awesome work.

Koepi
3rd November 2002, 19:24
Searching for

modulated quant test

in the XviD forum brings up this thread in 7th place:

http://forum.doom9.org/showthread.php?s=&threadid=30454&highlight=modulated+quant+test


You might want to read it.

Regards,
Koepi

serbersan
4th November 2002, 00:08
Thanks Koepi, I haven't found it before.