Technically, it’s oh so very simple really. I’ll try to take some screenshots which may shed some light - but in short:
- The real trick is focusing on shapes, silhouettes, depth, and composition, not the details. Not even the lighting, as this is bit limited (due to the way shaders are built)
- Characters and hero props are reaalllyyy decimated in zbrush
- The textures are very simple flat colors, maybe subtle gradients
- The Voronoi texture. This is magic, and works great to add the polygonal/angular detail that stays sharp. Road surface, etc.
- There’s a bit tricky shader based on Shader to RGB node for this hard shading
- There’s some Volumetric Fog in there
- There’s a million transparent planes, with gradient multiplied by Voronoi again, to create atmosphere. Some Alpha blended, some Additive
- The models are often faked, stretched, exagerrated to get the dynamic feel without resorting to motionblur. I mean, I could, but I don’t think the ‘softness’ of real motionblur fits the style.
I like the ‘Painterly Photoshop’ term you used - that’s how it feels to work on those. That’s EEVEE, I can see it as it will finally look, and just move things around, work with silhouettes by moving those transparencies - it’s how I would work in 2d/photoshop during the comp/post phase. Now I’m doing it in 3d and it’s awesome.
Those animated GIFs are just an afterthought at the moment - but since they are so easy and quick to do (thanks EEVEE), why not? To be honest, I’m bit surprised this works so well with camera moves…