Modding, Mission Design, and Coding > The Modding Workshop
PBR/BNAO
Subspace Anomaly:
After playing around with bunch of tools, I must say that maybe none of this is even required, it takes just few minutes to setup chaiNNer to convert metal to specular so none of this is even as important as I thought it is.
On the positive side, outside of DXT5nm compression and it's unpacking ( which I am not sure how much it improves on quality or space and even then it feels a bit wrong to do it runtime in shader, even though it should be pretty fast math ) the bent normals do seem to work quite well.
As for the converting pre-PBR textures, the major hurdle seems to be sins of the past, mostly having lights and shadows in the diffuse maps.
In this example I've basically butchered all the colors when I tried to level it up with curves.
EatThePath:
Regarding old textures, yeah, a wide slice of the community textures have AO bakes rolled into their diffuse textures. It was a great idea at the time(I always hated it a little because of the inherent flaws but it did have merit). And before that, just manually highlighting areas to imply curves the polycounts didn't support and so forth. Some ships have source files floating around without the AO layer flattened around, if you ask the original artist/projects or other artists they might have passed them around to. I even have had limited luck doing new AO bakes and then obscene things with blender nodes to try to undo the bake in the texture files, but only on a handful of ships with few nooks and fairly bright base hull textures. Trying to do it to the MVPs Sathanas effectively made me stop doing 3d art for about six months :banghead:
And yeah on the DXT5nm packing not necessarily being worthwhile anymore. With BC7 dds compression being the current recommendation for high end assets, I don't think the packing would be worth implementing today if we were starting form scratch. Dropping a channel maybe, but not shuffling them. We're a bit stuck with it, though.
Navigation
[0] Message Index
[*] Previous page
Go to full version