[M4IF Discuss] hourly usage fee for MPEG4

[Craig Birkmaier]

At 3:31 PM +0100 2/22/02, AVARO Olivier FTRD/DIH/REN wrote:
>
>>  And there is the very real possibility that someone can develop a new
>>  codec technology that avoids infringement on the the essential MPEG-2
>>  patents, and thus avoid paying the "MPEG-2 tax."
>Please tell us how you would proceed to achieve this :-)
>cu,
>O.

You begin by closely examining the IP behind core technologies from 
two perspectives:
1. Legacy patents and prior art - in particular look at a time window 
where these legacy patents would  move into the public domain within 
the time window for finalization of 26L. Assuming two years, given 
the typical MPEG-2 process, this would suggest that you could use any 
IP patented earlier than say 1985.  Likewise, there could be some 
value in examining prior art that might cast doubt on patented 
techniques developed after 1984.
2. Look at the flood of patents that were created in the 1990-1993 
time frame that are essential to MPEG-2. Try to avoid infringing them 
with 26L.
A few areas with which I have a little MPEG experience ( not all 
related to IP issues):
Do not include any tools for the coding of interlace in the core 26L 
standard. If the folks with all of the IP for coding interlace insist 
on having these tools, segregate them into an INTERLACE profile and 
let them create a separate license for their use.
Consider extended color gamuts for higher quality profiles. This 
should include support for 10 bit or greater luminance 
representations and full bandwidth color channels. I'm not sure if 
this relates to any IP issues with 4:2:0 or 4:2:2, but it is very 
important if we want to deliver high quality content to displays that 
routinely work in the RBG color space.
Avoid any attempts to define formats or PROFILES that are optimized 
for existing formats. Choose logical performance levels WITHOUT huge 
gaps like the one that exists between MPEG-2 MP   ML and MP   HL.
Take the time to fully document what conformance to the standard 
requires, especially with respect to maximization of utility of a 
performance level. For example,  you are NOT conformant if you create 
a downstream specification that imposes format restrictions on a 26L 
profile - and thus are not entitled to claim that you are using the 
26L standard.
Don't forget conformance testing, especially for reserved extensions. 
If an implementation is broken by the use of a conformant extension 
then the right to claim conformance with 26L should be revoked.
And there is one simple way to proceed if you are serious about a 
royalty free standard:
Impose a requirement that any essential IP be offered on a royalty 
free basis. Unfortunately I don't know if this is possible under ISO 
or ITU rules.
-- 
Regards
Craig Birkmaier
Pcube Labs