T 1989/17: Bump mapping math

T 1989/17: Bump mapping math

In T 1989/17, the Board considered the technical effect of a purely mathematical distinguishing feature in the context of a computer graphics rendering algorithm (“bump mapping”). While the case is not about AI or ML technology, and there is no question at the EPO that computer graphics is a technical field, the case has some valuable ammunition for arguing the technical effect of algorithmic efficiency in an application context. 

Bump mapping is a texture mapping technique in computer graphics for simulating bumps and wrinkles on a rendered surface. The details are not that relevant here, but it involves displacing surface normals so that the surface looks bumpy when lighting is applied. The surface normals are derived from filtered texture samples, and the invention here concerned how the filtering was done. While both the closest prior art and the claimed invention used B-Spline control points rather than Bezier control points in the filtering calculation (the technical details are secondary for our purposes), the invention did not convert the B-Spline control points to Bezier control points in this process. Instead, it used the B-Spline control points in a way that achieved the same effect. This was the distinction the Board focussed on.  

At first instance, the examining division found that since the same effect was achieved, and the methods differed only in details of the maths (not considered capable of being an invention per se), there was no technical effect, and no inventive step was acknowledged. The Board disagreed. The claimed method was more efficient in that it avoided conversion to Bezier control points, which provided a technical effect in the context of computer graphics processing. Nothing in the closest prior art suggested that anything other than Bezier control points could be used, and hence the appeal was allowed and the case remitted for grant.  

The Board’s reasoning is in line with the Guidelines for Examination for mathematical methods, which should have been applied by the examining division in the first place. The case, therefore, does not break new ground, but it is good to see this being applied by the Boards of Appeal (who are not bound by the Guidelines). This case goes on my list of useful cases for arguing about technical effects based on maths.  

Follow Alexander Korenberg’s AI musings

AI musings