The thing is, the characters need to be in the screen using the correct perspective. So making a simple 3d model, that’s basically only looked at from one or tw

The thing is, the characters need to be in the screen using the correct perspective. So making a simple 3d model, that’s basically only looked at from one or two perspectives that then allows to have a correctly rendered character, is perhaps less work than making an illustration and then creating a 3d mapping and occlusion model for the character. In the game, many of the scenes do use illustrated backgrounds. For example towards the end of the game, the caves. Here the calculation may have been the other way round (easier to do illustration).