Amazing article, well written and great starter for anyone who wants to learn how to use compact gradient texturing.
My only nitpick is that the title claims an in-engine performance analysis while only making surface level remarks about performance without any data to back it up. Some benchmark comparisons between gradient texturing with more vertices vs less vertices and higher resolution textures or even with lighting are needed to prove the real benefit, especially on mobile devices. I could not find any benchmarks or references to benchmarks in the paper either, although I don't speak Spanish so it's difficult to say.