nerd opened this issue on Jul 13, 2015 · 554 posts
pumeco posted Wed, 15 July 2015 at 8:07 AM
@Nerd
Just wanted to say, in addition to Baggins technical post, please do not underestimate the importance of simplification when you merge both engines into the same node system. I mentioned colour-coding earlier, but actually, there's a better way than that.
There needs to be a switch at the top of the Material Room where users can switch between CPU-NODES and GPU-NODES, because that way, no one will ever have to wonder whether each node works with the GPU or not. If you don't implement that, the uptake of the system will be impaired due to outright confusion. There is nothing better than knowing that every node you add, will work in the mode you have chosen, so there absolutely needs to be a mode selection switch (if you do end up giving access to the GPU). It might sound trivial, but believe me, that is one design oversight you absolutely need to avoid - there MUST be a switch to show only the relevant nodes for the renderer you're working with, otherwise, users won't know whether it is failing due to their own node structures or mode incompatibility.
There absolutely MUST be a switch.