My only concern with this is people expecting PBR assets made for MGE XE to look the same in OpenMW, which is not necessarily the case. I would be afraid of something similar to the whole normal map nonsense happening with PBR. I know that was kind of a special case, but as we've discussed before, just because something is made using PBR workflow doesn't mean it will appear the same in different applications. (Still, weird PBR assets are better than none, and the MGE implementation may be very good, I don't know.)AnyOldName3 wrote: ↑13 Sep 2019, 17:46 MGE XE has a PBR prototype, and a few things have been made for it. Once that settles down, it becomes more viable to start making similar changes in OpenMW.
I've been thinking a lot about this, and I think that this is inevitable. But the more we can guide it, the better the results will be. I have lots of thoughts on this so I think I'll bring them up in another thread. I do think the sooner we restart this the better. Just recently we had someone volunteer to start working on this as they didn't feel comfortable contributing in other ways. It would be nice to at least have guidelines to give people when they pop up and seem interested.AnyOldName3 wrote: ↑13 Sep 2019, 17:47 Also, if we ever resuscitate the MOAR project, we can make PBR-friendly materials a requirement for everything submitted to that.
The neural network being used in the linked example is an interesting place to start, but as with so many things, it seems better to simply start from scratch. (He seems to be having fun though, so more power to him)
I suspect it's more an issue of maintenance, but that seems like a good point. It may be worth starting on the gamma correct lighting if for no other reason than to start experimenting?