The Federal Circuit has upheld the patent-ineligibility under 35 U.S.C. § 101 of all claims of U.S. Patent No. 8,614,710, directed to “a method for deriving a pixel color in a graphic image.” Coffelt v. NVIDIA Corp., No. 2017-1119 (Fed. Cir. March 15, 2017) (per curiam and non-precedential; panel was Chief Judge Prost and Judges Bryson and Wallach). Although the court clearly did not regard this as a close case, it is worth noting for the court’s acknowledgement that the claims of the invalidated patent may recite a novel algorithm – but that novelty was not enough to make the claims patent-eligible.
Claim 1 of the ’710 patent recites:
A method for deriving a pixel color comprising the steps of:
a computer calculating a first position vector for a geometric graphic object;
a computer calculating a particular steradian region of space;
a computer calculating a particular steradian radius of said steradian region of space;
a computer calculating that said first position vector is located in said particular steradian region of space;
a computer calculating a second position vector for a geometric graphic object;
a computer calculating that said second position vector is located in said particular steradian region of space;
a computer calculating a length of said first position vector;
a computer calculating a length of said second position vector;
a computer comparing said first length to said second length;
for a first pixel, a computer deriving a pixel color for said first position vector from a result of said length comparison;
for a second pixel, a computer deriving a pixel color for said second position vector from a result of said length comparison.
The district court had granted the defendants’ motion to dismiss, finding that the claims of the ’710 patent recite an “abstract mathematical algorithm for calculating and comparing regions in space.” Of particular moment was that the claims could be performed by a human as mental steps and with a pen and paper.
The Federal Circuit agreed with the district court. The recitation of “calculating a particular steradian region of space” amounted to “nothing more than a mathematical algorithm that could be implemented using a pen and paper.” The ’710 patent claims thus failed the first (abstract idea) prong of the Mayo/Alice test.
Further, the claims failed the second (transformative inventive concept) prong of the Mayo/Alice test. Undisputedly, the claims could be implemented with a generic computer. Coffelt, the pro se patent owner, argued that the prior art, unlike his claims, did not allow for realistic renderings of 3D shadows. No matter, said the court:
The novelty of the algorithm, however, does not determine whether the claim recites an inventive concept. Instead, the inventive concept must “transform” the patent-ineligible algorithm into a “patent-eligible application” of the algorithm, and do so by more than merely implementing the algorithm on a generic computer. Alice, 134 S. Ct. at 2355. The asserted claims fail to do so here.
Plenty of claims “implemented on a generic computer” survive patent-eligibility challenges – see Enfish, LLC v. Microsoft Corp., for a prominent example. But this case is far from the first to invalidate patent claims under Section 101 on the ground that claims directed to mental steps are not patent-eligible. Like those cases, the distinguishing feature of this case is that it involved patent claims that recite a method that uses a computer, but does not, at least in theory, require one. Claims meeting this criterion are highly, if not fatally, suspect under Alice.