man/man1/bsdf2klems.1

.\" RCSid $Id: bsdf2klems.1,v 1.2 2013/08/02 20:56:19 greg Exp $
.TH BSDF2KLEMS 1 4/24/2013 RADIANCE
.SH NAME
bsdf2klems - generate XML Klems matrix description of a BSDF
.SH SYNOPSIS
.B bsdf2klems
[
.B "\-n spp"
][
.B "\-h|\-q"
][
.B "\-l maxlobes"
]
[
.B "bsdf.sir .."
]
.br
or
.br
.B bsdf2klems
[
.B "\-n spp"
][
.B "\-h|\-q"
]
.B bsdf_in.xml
.br
or
.br
.B bsdf2klems
[
.B "\-n spp"
][
.B "\-h|\-q"
][
.B "{+|-}forward"
][
.B "{+|-}backward"
][
.B "\-e expr"
][
.B "\-f file"
]
.B bsdf_func
.SH DESCRIPTION
.I Bsdf2klems
produces a Klems matrix representation of a
bidirectional scattering distribution function (BSDF)
based on an intermediate representation (in the first form),
an input XML representation (in the second form),
or a functional description (in the third form).
A complete XML description is written to the standard output,
which is normally redirected to a file.
.PP
The Klems matrix representation divides the input and output
hemisphere into a default 145 patches.
The
.I \-h
option may be used to reduce this number to 73 patches per hemisphere.
The
.I \-q
option may be used to reduce this number to 41 patches.
Neither option is recommended unless the distribution is known to
be approximately diffuse.
The
.I \-l
option may be used to specify the maximum number of lobes in any
interpolated radial basis function.
The default value is 15000, which generally keeps the interpolation tractable.
Setting the value to 0 turns off this limit.
.PP
Normally, multiple samples are taken from random points on each input
and output patch to improve accuracy.
The number of samples to take for each input-output patch pair may
be controlled using the
.I \-n
option, which defaults to 256.
.PP
The first invocation form takes a intermediate scattering representation
as produced by
.I pabopto2bsdf(1)
or similar, and produces a Klems representation with as many
components as there are independent input distributions.
Each intermediate scattering file contains one of
the four components, and if the first component
is isotropic, all components must be isotropic.
A similar rule holds for anisotropic inputs.
Only the center of each incident patches is sampled, due to
the time required to interpolate incident positions.
.PP
In the second invocation form, an input XML representation
is resampled to produce the desired Klems matrix representation.
This is primarily used to convert a tensor tree representation
into a matrix for annual daylighting simulations.
Any components in the input are reproduced on output, and inline
geometric descriptions are passed unchanged.
.PP
In the third invocation form,
.I bsdf2klems
takes a functional specification of a BSDF.
The named function should accept 6 parameters corresponding to the
normalized incident and exiting vectors, respectively.
By convention, these vectors point away from the surface, and a positive
Z-component corresponds to the front side.
The Y-component corresponds to the "up" orientation of the surface,
as specified in the eventual scene description that references the XML
output.
If the function only takes 3 parameters, then the variables "Dx", "Dy",
and "Dz" will be assigned to the reverse of the outgoing direction at
each evaluation.
(I.e., the vector will point into the surface and
Dz will be negative on the front side.)\0
This simplifies conversion of functional BSDF specifications using the
legacy material primitives "plasfunc", "metfunc", and "transfunc".
.PP
The function is defined by one or more
.I \-e
and
.I \-f
options, and should obey both Helmholtz reciprocity and
integrate to less than 1 over each projected incident hemisphere
for energy conservation.
.PP
Similar to the
.I genBSDF(1)
command,
the
.I \+backward
option (default) specifies that rays arriving from the front side of
the surface will be tested for reflection and transmission.
If both forward and backward (front and back) distributions are needed, the
.I \+forward
option may be given.
To turn off the backward components, use the
.I \-backward
option.
Computing both incident hemispheres takes about twice as long as one, but
is recommended when rays will be impinging from either side.
.SH EXAMPLE
To take two components of an intermediate BSDF representation and create
a full Klems matrix representation:
.IP "" .2i
bsdf2klems transmitted.sir reflected.sir > combined.xml
.PP
To reduce a tensor tree representation into a half-Klems matrix representation:
.IP "" .2i
bsdf2klems -h ttree.xml > klems_half.xml
.PP
To create a low-res BSDF corresponding to a one-sided,
isotropic Phong distribution:
.IP "" .2i
bsdf2klems -e 'phong(ix,iy,iz,ox,oy,oz) = if(iz, .1+((iz+oz)/sqrt((ix+ox)^2+(iy+oy)^2+(iz+oz)^2))^50, 0)' phong > phong.xml
.SH AUTHOR
Greg Ward
.SH "SEE ALSO"
bsdf2ttree(1), dctimestep(1), icalc(1), gendaymtx(1), genklemsamp(1),
genskyvec(1), mkillum(1), genBSDF(1), pkgBSDF(1), rcontrib(1), rtrace(1)
Revision:	1.3
Committed:	Thu Sep 26 17:05:00 2013 UTC (12 years, 1 month ago) by greg
Branch:	MAIN
Changes since 1.2:	+11 -3 lines
Log Message:	Added -l option to limit maximum number of RBF lobes for interpolation
#	User	Rev	Content
1	greg	1.3	.\" RCSid $Id: bsdf2klems.1,v 1.2 2013/08/02 20:56:19 greg Exp $
2	greg	1.1	.TH BSDF2KLEMS 1 4/24/2013 RADIANCE
3			.SH NAME
4			bsdf2klems - generate XML Klems matrix description of a BSDF
5			.SH SYNOPSIS
6			.B bsdf2klems
7			[
8			.B "\-n spp"
9			][
10			.B "\-h\|\-q"
11	greg	1.3	][
12			.B "\-l maxlobes"
13	greg	1.1	]
14			[
15			.B "bsdf.sir .."
16			]
17			.br
18			or
19			.br
20			.B bsdf2klems
21			[
22			.B "\-n spp"
23			][
24			.B "\-h\|\-q"
25			]
26			.B bsdf_in.xml
27			.br
28			or
29			.br
30			.B bsdf2klems
31			[
32			.B "\-n spp"
33			][
34			.B "\-h\|\-q"
35			][
36			.B "{+\|-}forward"
37			][
38			.B "{+\|-}backward"
39			][
40	greg	1.3	.B "\-e expr"
41	greg	1.1	][
42	greg	1.3	.B "\-f file"
43	greg	1.1	]
44			.B bsdf_func
45			.SH DESCRIPTION
46			.I Bsdf2klems
47			produces a Klems matrix representation of a
48			bidirectional scattering distribution function (BSDF)
49			based on an intermediate representation (in the first form),
50			an input XML representation (in the second form),
51			or a functional description (in the third form).
52			A complete XML description is written to the standard output,
53			which is normally redirected to a file.
54			.PP
55			The Klems matrix representation divides the input and output
56			hemisphere into a default 145 patches.
57			The
58			.I \-h
59			option may be used to reduce this number to 73 patches per hemisphere.
60			The
61			.I \-q
62			option may be used to reduce this number to 41 patches.
63			Neither option is recommended unless the distribution is known to
64			be approximately diffuse.
65	greg	1.3	The
66			.I \-l
67			option may be used to specify the maximum number of lobes in any
68			interpolated radial basis function.
69			The default value is 15000, which generally keeps the interpolation tractable.
70			Setting the value to 0 turns off this limit.
71	greg	1.1	.PP
72			Normally, multiple samples are taken from random points on each input
73			and output patch to improve accuracy.
74			The number of samples to take for each input-output patch pair may
75			be controlled using the
76			.I \-n
77			option, which defaults to 256.
78			.PP
79			The first invocation form takes a intermediate scattering representation
80			as produced by
81			.I pabopto2bsdf(1)
82			or similar, and produces a Klems representation with as many
83			components as there are independent input distributions.
84			Each intermediate scattering file contains one of
85			the four components, and if the first component
86			is isotropic, all components must be isotropic.
87			A similar rule holds for anisotropic inputs.
88			Only the center of each incident patches is sampled, due to
89			the time required to interpolate incident positions.
90			.PP
91			In the second invocation form, an input XML representation
92			is resampled to produce the desired Klems matrix representation.
93			This is primarily used to convert a tensor tree representation
94			into a matrix for annual daylighting simulations.
95			Any components in the input are reproduced on output, and inline
96			geometric descriptions are passed unchanged.
97			.PP
98			In the third invocation form,
99			.I bsdf2klems
100			takes a functional specification of a BSDF.
101	greg	1.2	The named function should accept 6 parameters corresponding to the
102	greg	1.1	normalized incident and exiting vectors, respectively.
103			By convention, these vectors point away from the surface, and a positive
104			Z-component corresponds to the front side.
105			The Y-component corresponds to the "up" orientation of the surface,
106			as specified in the eventual scene description that references the XML
107			output.
108	greg	1.2	If the function only takes 3 parameters, then the variables "Dx", "Dy",
109			and "Dz" will be assigned to the reverse of the outgoing direction at
110			each evaluation.
111			(I.e., the vector will point into the surface and
112			Dz will be negative on the front side.)\0
113			This simplifies conversion of functional BSDF specifications using the
114			legacy material primitives "plasfunc", "metfunc", and "transfunc".
115	greg	1.1	.PP
116			The function is defined by one or more
117			.I \-e
118			and
119			.I \-f
120			options, and should obey both Helmholtz reciprocity and
121			integrate to less than 1 over each projected incident hemisphere
122			for energy conservation.
123			.PP
124			Similar to the
125			.I genBSDF(1)
126			command,
127			the
128			.I \+backward
129			option (default) specifies that rays arriving from the front side of
130			the surface will be tested for reflection and transmission.
131			If both forward and backward (front and back) distributions are needed, the
132			.I \+forward
133			option may be given.
134			To turn off the backward components, use the
135			.I \-backward
136			option.
137			Computing both incident hemispheres takes about twice as long as one, but
138			is recommended when rays will be impinging from either side.
139			.SH EXAMPLE
140			To take two components of an intermediate BSDF representation and create
141			a full Klems matrix representation:
142			.IP "" .2i
143			bsdf2klems transmitted.sir reflected.sir > combined.xml
144			.PP
145			To reduce a tensor tree representation into a half-Klems matrix representation:
146			.IP "" .2i
147			bsdf2klems -h ttree.xml > klems_half.xml
148			.PP
149			To create a low-res BSDF corresponding to a one-sided,
150			isotropic Phong distribution:
151			.IP "" .2i
152			bsdf2klems -e 'phong(ix,iy,iz,ox,oy,oz) = if(iz, .1+((iz+oz)/sqrt((ix+ox)^2+(iy+oy)^2+(iz+oz)^2))^50, 0)' phong > phong.xml
153			.SH AUTHOR
154			Greg Ward
155			.SH "SEE ALSO"
156			bsdf2ttree(1), dctimestep(1), icalc(1), gendaymtx(1), genklemsamp(1),
157			genskyvec(1), mkillum(1), genBSDF(1), pkgBSDF(1), rcontrib(1), rtrace(1)