Hi,
Is there a consenus on how much of maps baked from high-res source should be direct from the render or done on the game model itself?
I can see the advantages of creating most of it from the high-res model (which means that the high-res model would require some mapping as well).
You could then mess around with the UVs on your game model without worrying about chopping up the diffuse texture, as it would just get baked out the same time as the normal map.
I've tended to use different diffuse passes (e.g. a skylight pass, procedurals etc. to get the majority of the final diffuse pass) but wondered how other people approach it.
Thanks,
- Steve
Replies
therefore i try to stay away from baking whenever possible. often i would hand paint height maps to use for bump/normal-passes just to avoid fiddling too much on a model with insane resolution. seems much more efficient to me.
but i agree that baking out more than just your standard normal map can be handy at times, but i try to restrict it's use only to tackle very specific problem areas to avoid working too much on the highpoly itself.
I use a lightdown setup for my lighting information.
Generally with the high-res model I map it as a medium poly model, then just turbosmooth at baking time.
I think there a couple of reasons for me preferring to get as much from the original model.
1. Mapping which suits normal maps (like pelt mapping) I find a lot harder to paint flat in a 2D package due to distortion.
2. Using procedurals will sort out seams for free - and maps like Darksim's free Simbiont maps are pretty flexible.
Also it's handy to have a nicely textured high-res model
Cheers
Steve