Hello!

This goes to kwag and Dialhot

Could I upload a KDVD template that I created on cce-sp 2.70.02.01 , for you guys verify if everything is fine ?

best regards

sparskter

I'm not sure I will upload it because I don't want to screw my own settings (but I can be in a good mood, who knows ? ) but there is no problem.

Well, here is the download link:

http://rapidshare.de/files/6214260/K...02.01.zip.html

EDIT: link fixed! Do not forget to apply Dialhot recommendations!!

If anyone can verify if everything is fine

So, if everything is accordingly to the KDVD specs, can I assume that Half-KDVD and KSVCD only use less bitrate ?

best regards

sparskter

Little things to change (not important, just advices) :

1/ you should do anamorphic DVD insteed of 4:3 ones, so select 16:9
2/ you can uncheck the "Video information file" box. It is useless in 1pass
3/ you should not use Q factor above 30. And max bitrate can be setted to 6000 for general jobs. 8000 is needed for few sources indeed.
4/ I suggest you to name the matrix and save it. You will avoid to retype it is you loose your template !
(to save a matrix, change any values to ungrey the "save" button)
5/ In the picture quality tab, set the DC precision to 9

Hello Dialhot!

My source will be mostly 640x480 animes. So should be 16:9 or 4:3 ? And if my source is "wide screen" should set it to 16:9?


I was thinking this was about the same as CQ in tmpgenc. So it is NOT. The less the value of Q factor the more quality right ? (and bigger the file...)
good tip!
I would like to know more about DC precision. What is it´s influence on my video encode ? If I change it to 8 in tmpgenc, the encode gets a lot faster!!

Thanks for the review

USe anamorphic only if the source is anamorphic also. Only DVD are anamorphic.

Yes it is.

The speed should not be so different except if you have a "slow" PC.
DC precision is the number of digits after the dot used during mathematical operation. Using a bigger precision gives files a little smaller but the picture is a little bit softer. For KVCD we use DC=8 but for DVD DC=9 is a better choice IMHO.

Sorry posting again but....

About Q-Factor it is not still clear for me

You said: "Yes it is"

It is... what ? -LOL-

So, 1 question:
I am saying this because you told me to use a value at maximum 30 (so, the less, the better at cost of file space.) And in tmpgenc:



is the vice-versa: the higher, the better at cost of space.

thanks ! sorry any inconvenience !

Look at your post. Your only question was : right ?
"Yes it is" was the answer to that question.

In a longer version : yes the Q factor works oppositely to CQ value. The lower the Q factor, the better the quality (the lesser the compression in fact). And of course, the filesize is bigger.

I missed one important point....

I was browsing the windows registry and:



shoudn´t the GOP length be 18 for NTSC encodes ?

thanks once again

18 is the MAX allowed for NTSC by the standard but CCE can't produce GOP longer than 15.

Phil Is that true even with the latest version of CCE?

If so why do you think they did not increase the max GOP range?

__________________
Regards.

Michael.

It is true for any version.

They do not care about the DVD Spec, they did a MPEG encoder. In the last release you have a box to check to restrict what is produced to what allows DVD spec (video resolution and bitrates for instance), but not to extend what CCE does to the limit of DVD specs. They just didn't think about the GOP length.

Edit: I will see if this is possible to tweak it directly into the registry. Perhaps the limitation is only in the GUI of the tool, but you can bypass it !

I'd rather know why the DVD limit is different between PAL and NTSC. There is no technical reason I can imagine.

__________________
Regards.

Michael.

I might be wrong, but from experience encoding a 23.976 video with 18 GOP will give more than 36 fields GOP (or 18 frames) at playback after the pulldown applied as the pulldown will mark some fields to be repeated. That is, I think, why the GOP size in CCE doesn't exceed 15 GOP.

If they thought like this this is weird : fields are duplicated after decompressing of course. This does not enter into account in the GOP length (that has a meaning only in compressed context).

BTW 15*30/24 = 18.75, that is still bigger than 18.

Well, maybe, I really don't know. The fact is that I had a warning about that from an authoring programm that require files to be pulldowned before import. And about 15 GOP - well, they must think also about the PAL users