I thought the only two 100% compute games, Dreams and Claybook, which are beautiful, but appear limited in scope, weren't released yet?
they are both releasing for this gen, so you don't need next gen hardware to have whole compute pipelines.
These type of games are limited in scope because of the amount of R&D to make everything work for these types of engines. I'd say it's an issue surrounding tools and being the first to pioneer many methods required by the new compute only engine vs a lot of techniques that are tried tested and true on rasterization.
If you look at why games in the AAA space evolve so slowly, it's because tools need to evolve as well. Rewriting your renderer to be 100% compute based means tossing/rewriting all your tools, and that's detrimental to the success of any game, especially when it takes years to craft and build a large open world game, you can't have your world makers stuck for years waiting to for tools to build.
So we only get these types of games in the indie scene. And it will be a long time until we see 100% compute engines in the AAA scene. It's a slow transition, but they will become more compute over time, they just can't do a hard cut like smaller studios can, since smaller studios are more likely to rely on abstract graphics, where AAA artwork is fairly different.
I barely know anything about SDF, but
if this is how they have been creating content without some sort of importer/conversion tool that's painful.
So in some AAA studio they're probably using some 3D modelling tool that operates in polygons, and that model has importers to import into the game. They'll need to have some way, either through polygons, or perhaps some sculpting tool, or a vector based 3d modeller, or some other tool to import into a SDF object.
I suspect it's doable, but I don't suspect it's easy. We're also looking at a whole workflow adjustment etc.
I think these are the realistic reasons we don't see SDF in AAA yet.
