It appears that the Traube Diacrhome process that's referenced in this article is similar (if not identical) to the dye mordant process that R Paul alluded to and that F.E. Ives utilized.
This article gives a very clear idea of the necessary type of dyes and notes 3 possible characteristics. One is that the dye simply doesn't color gelatin period, the second that it stains only the shadows/unexposed portions and leaves the highlights clear, which is what we'd need if we were working with planographic (or Pinatype) matrices, and the third is that the dye stains the whole matrix equally. This last characteristic is probably what we see with most imbibition dyes and this kind of dye would only work in a relief matrix (hot water etched) type of system.