I want fully dynamic shadows for my game. The characters are already very low poly (a character mesh can have 500-600 vertices with 900 faces). But at the same time I want to squeeze as much performance as I can. Is it reasonable to use 2 skeletal meshes, one ordinary (non-shadow casting), and another to cast shadows? As far as I can understand geometry complexity of an object casting shadows plays a big role.
And another question, a bit more general, is it possible to make a game with fully dynamic shadows that will run smoothly on low spec devices? I know there were really old games utilizing dynamic shadows like severance, in a limited way, but still.
Hi,
Is it reasonable to use 2 skeletal meshes, one ordinary (non-shadow casting), and another to cast shadows?
You could use capsule shadows instead, might/would likely be cheaper (and is already build in).
And another question, a bit more general, is it possible to make a game with fully dynamic shadows that will run smoothly on low spec devices? I know there were really old games utilizing dynamic shadows like severance, in a limited way, but still.
You could create a testmap, package it and then see how it performs on the low spec device you want to run it on. Maybe then also use ue4 (and maybe a ue4 version before chaos replaced nvidia physx) instead of ue5…