RE: [Algorithms] SH data falloff..
Brought to you by:
vexxed72
From: Tom F. <tom...@ee...> - 2004-01-17 16:52:07
|
So the only way this will work is if you re-create the irradiance SH at = each vertex, rather than at each object centre. Doing this will take into = account all the incident angle and intensity changes. TomF. > -----Original Message----- > From: gda...@li...=20 > [mailto:gda...@li...] On=20 > Behalf Of Andreas Tadic > Sent: 17 January 2004 16:32 > To: gda...@li... > Subject: SV: [Algorithms] SH data falloff.. >=20 >=20 > Indeed. The thing I'm doing here is offline - after the shprt sim. > Calculating luminance for all static scene objects at one light at the > time (and the storing the results). > So here the # of lights is "infinite".. > Works really well.. Except of course the "seam" :( >=20 >=20 > -----Ursprungligt meddelande----- > Fr=E5n: gda...@li... > [mailto:gda...@li...] F=F6r Tom = Forsyth > Skickat: den 17 januari 2004 16:53 > Till: gda...@li... > =C4mne: RE: [Algorithms] SH data falloff.. >=20 >=20 > If you're doing all this with a single light, then maybe you=20 > can correct > in some way. Thing is, if you're only doing one light at a time, then > using an SH doesn't seem to be the right answer - it's=20 > overcomplicating > things. The real win with using SHs is that it doesn't matter how many > lights you use, the cost is the same. >=20 > TomF. >=20 >=20 > > -----Original Message----- > > From: gda...@li... > > [mailto:gda...@li...] On=20 > > Behalf Of Andreas Tadic > > Sent: 17 January 2004 15:43 > > To: gda...@li... > > Subject: SV: [Algorithms] SH data falloff.. > >=20 > >=20 > >=20 > > Hmm. I see what you mean. > > I would have assumed that once one knows the lumination of the=20 > > centerpoint of the object (that the prt sim is doing) one could use=20 > > that formula to modify the luminance of each vector in the object. > > (And then use the greyscale luminance in ones point-light=20 > calculation) > > Obviously not at simple as that.. > >=20 > > -----Ursprungligt meddelande----- > > Fr=E5n: gda...@li... > > [mailto:gda...@li...] F=F6r=20 > Tom Forsyth > > Skickat: den 17 januari 2004 14:40 > > Till: gda...@li... > > =C4mne: RE: [Algorithms] SH data falloff.. > >=20 > >=20 > > Agreed. Abutting objects are a real pain when trying to do=20 > any sort of >=20 > > light simplification - very hard to get rid of the seams. Computing=20 > > the SHs where they meet and interpolating is the only=20 > solution I know=20 > > (not necessarilly along the corners, just so that along the=20 > line where >=20 > > they meet the interpolation is the same). But of course it's=20 > > moderately expensive, and in this case you need to=20 > interpolate between > > four sets of > > SHs per object. > >=20 > > The other option I thought of was to mark the vertices as=20 > "edge verts" >=20 > > and light them with the standard point-light illumination. This=20 > > ensures they are consistent across seams. Only vertices for=20 > the rest=20 > > of the object are lit using an SH. The problem is that the=20 > lighting is > > inconsistent across the surface of each object, and also that=20 > > it doesn't > > actually lessen the workload (which is the whole point of=20 > > using SHs for > > lighting) unless you have conditional execution in your=20 > > vertex shaders, > > which is only in pretty fancy hardware right now. > >=20 > > TomF. > >=20 > > > -----Original Message----- > > > From: gda...@li... > > > [mailto:gda...@li...] On > > > Behalf Of Ignacio Casta=F1o > > > Sent: 17 January 2004 13:22 > > > To: gda...@li... > > > Subject: Re: [Algorithms] SH data falloff.. > > >=20 > > >=20 > > > Andreas Tadic wrote: > > > > Each object is at a distance from the light, pivot in the > > > center of each > > > > object. What were attempting is to smooth out the=20 > vertices on each >=20 > > > > object to get be able to post-process a pointlight using > > > this data to > > > > control the pointlights luminace with, in effect being > > able to have > > > > pointlights on SH objects. Since SH objects have their > > > luminance based > > > > on the center of the object (it's position) they get > > > "tiled" as shown here: > > > > _http://www.customred.com/sh/rawsh.jpg_ (rawsh.cpp snippet) > > >=20 > > > I think that you should compute the SHs on the corners of=20 > the tiles=20 > > > instead of the middle, and interpolate the coefficients. > > >=20 > > > You last shots look good, though. How are you doing that? > > >=20 > > > -- > > > Ignacio Casta=F1o > > > cas...@ya... > >=20 > >=20 > >=20 > >=20 > >=20 > > ------------------------------------------------------- > > The SF.Net email is sponsored by EclipseCon 2004 > > Premiere Conference on Open Tools Development and=20 > Integration See the=20 > > breadth of Eclipse activity. February 3-5 in Anaheim, CA.=20 > > http://www.eclipsecon.org/osdn=20 > > _______________________________________________ > > GDAlgorithms-list mailing list=20 > GDA...@li... > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > Archives: > > http://sourceforge.net/mailarchive/forum.php?forum_ida88 > >=20 > >=20 > >=20 > >=20 > > ------------------------------------------------------- > > The SF.Net email is sponsored by EclipseCon 2004 > > Premiere Conference on Open Tools Development and=20 > Integration See the=20 > > breadth of Eclipse activity. February 3-5 in Anaheim, CA.=20 > > http://www.eclipsecon.org/osdn=20 > > _______________________________________________ > > GDAlgorithms-list mailing list=20 > GDA...@li... > > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > > Archives: > > http://sourceforge.net/mailarchive/forum.php?forum_ida88 > >=20 >=20 >=20 >=20 > ------------------------------------------------------- > The SF.Net email is sponsored by EclipseCon 2004 > Premiere Conference on Open Tools Development and Integration See the > breadth of Eclipse activity. February 3-5 in Anaheim, CA. > http://www.eclipsecon.org/osdn > _______________________________________________ > GDAlgorithms-list mailing list GDA...@li... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: http://sourceforge.net/mailarchive/forum.php?forum_ida88 >=20 >=20 >=20 >=20 > ------------------------------------------------------- > The SF.Net email is sponsored by EclipseCon 2004 > Premiere Conference on Open Tools Development and Integration > See the breadth of Eclipse activity. February 3-5 in Anaheim, CA. > http://www.eclipsecon.org/osdn > _______________________________________________ > GDAlgorithms-list mailing list > GDA...@li... > https://lists.sourceforge.net/lists/listinfo/gdalgorithms-list > Archives: > http://sourceforge.net/mailarchive/forum.php?forum_ida88 >=20 |