12 Sep
2003
12 Sep
'03
8:35 a.m.
Michael Niedermayer wrote:
I don't really understand mbd=2, but I'd recommend doing some testing.
it chooses the encoding which minimizes sum_of_squared_errors + number_of_bits*lambda
It might only improve psnr with 2pass encoding.
no, its always better
So, can anybody explain the relationship between vqscale, vqmin, vqmax, mbqmin, and mbqmax? Is specifying vqscale the same as specifying vqmin and vqmax with the same value? Does mbqmin limit minimum quantizers even if vqmin is lower? Is there ever a good reason to change mbqmin and mbqmax? Jonathan Rogers