Re: [Algorithms] Tangent space lighting problem using radiosity normal mapping

[Aras P. <ne...@gm...>]

> > Just a thought - maybe it's all "ok", considering that where you have
> > different tangent bases, you also have different "radiosity lighting"
> > bases. At this discontinuity, the incoming light signal (the one that
> > eventually is stored in 3 lightmaps) gets approximated differently and
> > there you have your result.
>=20
> Hmm... I think it's not ok, because the directional basis used in
> "radiosity normal mapping" are not rotationally invariant. It would be
> ok though, if lighting basis was in a fixed space and you translated
> the normals to that space instead, but that's probably not the case
> here.

That's what I was trying to say - the results are "ok" as far as the
lighting algorithm is concerned. That is, you should get lighting
discontinuities where you have lighting space discontinuities. This of
course doesn't solve the problem "how do I get rid of them"...


--=20
Aras 'NeARAZ' Pranckevicius
http://nesnausk.org/nearaz | http://nearaz.blogspot.com