mmp
Equivalent of Kt and Kr in dielectric material?
I am trying to achieve the effect of colored glass in v4. Information I have found online (e.g. here) points me to the Kt property of the glass material in v3. "color Kt" [0.7 1 0.7 ]
v4's dielectric material does not have the Kt ad Kr properties which were available for glass in v3. If I try and start from a v3 file with glass, the --upgrade function converts "color Kt" [0.7 1 0.7 ] to "rgb Kt" [0.7 1 0.7 ]. The result, however, is rejected by v4 with the error ""Kr": unused parameter".
What is the right way to achieve colored glass in v4?
In pbrt-v4 I believe the approach to take is to use a homogeneous Medium and set the appropriate absorption values.
Left sphere is a diffuse transmission, right sphere is a default dielectric with a interior medium that absorbs everything but red.
PixelFilter "gaussian" "float xradius" [ 2 ] "float yradius" [ 2 ]
Sampler "zsobol" "integer pixelsamples" [ 8192 ]
Integrator "volpath" "integer maxdepth" [ 5 ]
Accelerator "bvh"
Film "gbuffer"
"string filename" [ "pbrt.exr" ]
"integer xresolution" [ 1280 ]
"integer yresolution" [ 720 ]
Transform [ 0.707106781186547462 -0.353553390593273731 -0.612372435695794581 0.0
0 0.866025403784438597 -0.5 0.0
-0.707106781186547573 -0.353553390593273897 -0.612372435695794359 0.0
0.0 0.0 4.0 1.0 ]
Camera "perspective" "float fov" [ 45 ] "float screenwindow" [ -1 1 -0.5625 0.5625 ]
WorldBegin
AttributeBegin
Translate -2 3 0
Rotate -50 0 0 1
Rotate -90 0 1 0
AreaLightSource "diffuse"
"float scale" [ 200 ]
"bool twosided" [ false ]
AttributeBegin
Rotate 180 1 0 0
Shape "disk" "float radius" [ 0.25 ]
AttributeEnd
AttributeEnd
Texture "/mat/checkerboard" "spectrum" "checkerboard"
"rgb tex1" [ 0.75 0.75 0.75 ]
"rgb tex2" [ 0.3 0.3 0.3 ]
"float uscale" [ 300 ]
"float vscale" [ 300 ]
MakeNamedMaterial "/mat/ground"
"string type"
"diffuse" "texture reflectance" [ "/mat/checkerboard" ]
MakeNamedMaterial "/mat/red_diffuse"
"string type" "diffusetransmission"
"rgb transmittance" [ 1 0 0 ]
MakeNamedMaterial "/mat/glass" "string type" "dielectric"
"rgb eta" [ 1.33 1.33 1.33 ]
"float roughness" [ 0.001 ]
AttributeBegin
MakeNamedMedium "/mat/red_interior"
"string type" "homogeneous"
"float scale" [ 10 ]
"rgb sigma_a" [ 0.01 1.0 1.0 ]
"rgb sigma_s" [ 0.01 0.01 0.01 ]
AttributeEnd
AttributeBegin
NamedMaterial "/mat/ground"
AttributeBegin
ReverseOrientation
Shape "bilinearmesh"
"normal N" [ 0 1 0 0 1 0 0 1 0 0 1 0 ]
"point3 P" [ -50 0 -50 50 0 -50 -50 0 50 50 0 50 ]
"integer indices" [ 0 1 2 3 ]
AttributeEnd
AttributeEnd
AttributeBegin
Translate -0.25 0.5 0.25
Scale 0.2 0.2 0.2
NamedMaterial "/mat/red_diffuse"
AttributeBegin
Shape "sphere"
AttributeEnd
AttributeEnd
AttributeBegin
Translate 0.25 0.5 -0.25
Scale 0.2 0.2 0.2
NamedMaterial "/mat/glass"
MediumInterface "/mat/red_interior" ""
AttributeBegin
Shape "sphere"
AttributeEnd
AttributeEnd
I used the GPU (volume pathtracer), then ran it through imgtool's denoiser.
If I try and start from a v3 file with glass, the --upgrade function converts "color Kt" [0.7 1 0.7 ] to "rgb Kt" [0.7 1 0.7 ]. The result, however, is rejected by v4 with the error ""Kr": unused parameter".
Can you confirm this? If I make a simple scene that contains
Material "glass" "color Kt" [0.5 0.5 0.5]
I get a warning from pbrt --upgrade, "Warning: s.pbrt:37:0: Parameter is being removed when converting to "dielectric" material: "rgb Kt" [ 0.5 0.5 0.5 ]", and indeed, the converted pbrt file just has
Material "dielectric"
(And thanks to @shadeops for that nice alternative.)
The workaround from @shadeops works, but it doesn’t seem to produce exactly the same result.
Do I understand correctly that the homogeneous Medium simulates individual scattering events one by one, whereas Kt simply attenuates light exponentially along the path length — effectively accounting for an infinite number of scattering events?
pbrt-v4 (maybe it will converge eventually...):
LuxCoreRender:
