Apple's RAND licensing, prior art, and the W3C process

Dear W3C,

I would like to thank the W3C for extending the review period.

I stand by my previous comments (posted earlier to this list),
and I agree with most contributors to the review that, at the very
least, a radical revision of the working-draft is required to prevent
the W3C patent policy from extinguishing free software, and the W3C
with it.

I would like to add a data point regarding the SVG 1.0 spec that
might cast some light on the effectiveness, and perhaps the desirability,
of the RAND process.

At http://www.w3.org/2001/07/SVG10-IPR-statements.html there is a
document "SVG 1.0 Patent Statements". Out of 11 companies with patents,
only 3 were not willing to grant royalty-free licenses.

The first patent cited as being unavailable for RF-licensing for
SVG 1.0, Apple's US Patent 5379129, appears to be an attempt to patent
the technique of 'alpha blending' of images.

I found this of particular interest, as I have personal knowledge
that this alpha-blending technique was in wide use in the film and video
special-effects industry several years before the filing date of 1992.

For example, in Porter and Duff's 1984 paper 'Compositing Digital Images',
(cited below) they say (on page 256):

  "This is almost the well-known linear interpolation of
   foreground F with background B

   B' = (F x alpha) + B x (1 - alpha)"

Compare this with the wording of Apple's patent.

This appears to be prior art that puts the validity of Apple's patent
into question. There are also precedents for the other claims in the
patent regarding one-bit masking, soft masks and so forth. 

In light of this, I would ask the W3C to consider

a    if Apple's stance regarding licencing this patent under RAND terms
     advances the W3C's goals, and

b    if setting licence terms for patents of this sort is an appropriate
     use of W3C resources

c    whether this patent should have been subjected to review prior to being
     put forward for the RAND process
 
I would urge the W3C to adopt the following policies:

1    Insist on full disclosure of all possibly relevant patents by all W3C
     contributors at the earliest possible moment

2    Initiate a public review of prior art for all such patents before
     considering standardising any potentially infringing technologies
     unless a royalty-free licence has been granted by the patent owner.

     This review should be fully publicised, and adequate time given for
     responses.

3    Open up W3C membership to individual members with knowledge
     of relevant technologies, to allow the review to be as effective
     as possible.

Yours sincerely,

Neil Harris

-------------------------------------------------------------------------------------------------------

References:

   http://www.w3.org/2001/07/SVG10-IPR-statements.html

   US Patent 5379129:
   "Method for compositing a source and destination image
   using a mask image"
   Filed: May 8, 1992
   Granted: January 3, 1995
   Apple Computer, Cupertino, CA

    http://patft.uspto.gov/netahtml/srchnum.htm

   Porter Tom; and Duff, Tom: "Compositing Digital Images".
   Computer Graphics Vol. 18, No. 4 (July 1984) pp. 253 - 259.

   http://keithp.com/~keithp/porterduff/Page256.html

Received on Wednesday, 10 October 2001 15:59:58 UTC