What about spoilage ?
 

Hi all,

Sorry if I post in the wrong forum but I could'nt decide if this subject was better there or in KVCD template forum.

As some of you (all ? ;-)), I try to find how I can raise my CQ without modifying the length of the video encoded. I remember an old post of Sansgrip telling that with some spoilage, the samples were smaller. So I did some tests. And there are the results :

(I worked in VLBR, with min = 64kb/s, max = 1150kb/s and CQ = 70)

MPEG1 with P pict spoilage = 5 and B pict spoilage = 20
--> samples length = 1467 Ko

MPEG2 with P pict spoilage = 5 and B pict spoilage = 20
--> samples length = 1455 Ko

MPEG1 with P pict spoilage = 0 and B pict spoilage = 0
--> samples length = 1590 Ko

MPEG1 with P pict spoilage = 0 and B pict spoilage = 0
--> samples length = 1624 Ko


Conclusions:

1/ MPEG2 has better results than MPEG1 but the video looks worst. So the winner is MPEG1.

2/ with spoilage, I have a gain a 10% on the sample size !

The question is : shouldn't we use spoilage to make our KVCD ? :lightbulb: :|

I think that spoilage just shifts the CQ curve, it was probably discussed in that huge CQ vs. CQ_VBR thread.

Ditto :!:
It just "offsets" the CQ curve :wink:

-kwag