At 05:56 PM 1/18/00 , John Lazzaro wrote:
>the
>only interpolation method that is normative is linear interpolation
>(which is the default, or can be set explicitly with the global
>parameter "interp 1"). Setting "interp 2" results in a method
>the standard says should be "better than linear", and leaves it
>up to the implementator to define further.
John, just a small point. I think that linear is "interp 0" and
more than linear is "interp 1".
>I believe if you avoid everything above, you'll end up with a
>fully normative composition, that any compliant decoder should
>render identically as defined by the standard. Hmm, this would
>be a good feature for sfront to add -- a mode that warns people
>when they are doing something non-normative ...
I wouldn't say something non-normative, I'd say something with
non-normative implementation methods. A composition with
lopass or hipass *is normative*, simply not in a byte-exact way.
It will have to satisfy some other conformance criteria (in this
case it will be an SNR mask).
Stay tuned. I will do my best after the end of January to inform you
about Conformance (because FCD is still very uncomplete and
we need DIS, as I said this morning).
Best regards,
Giorgio
__________________________________________________________________
Giorgio ZOIA
Integrated Systems Laboratory - DE/LSI - EPFL
CH-1015 Lausanne - SWITZERLAND
Phone: + 41 21 693 69 79 E-mail: Giorgio.Zoia@epfl.ch
Fax: +41 21 693 46 63
__________________________________________________________________
This archive was generated by hypermail 2b29 : Mon Jan 28 2002 - 11:46:37 EST