Hello,
I am new to Radiance, have some questions and hope
that you can help me.
Here is an image to illustrate my questions:
About spotlights.
1.)
Does each point of e.g. a ring act as single spotlight when spotlight
is assigned to a surface? (upper left image)
2.)
When I assign a spotlight to a ring, how does the direction of the ring's normal
and the direction of the spotlight act together?
I would have expected the light spot to get elliptic when the spotlight is pointing
downwards but the ring has a different normal. (like the projected
area of the ring when looking in the direction of the light)
But the light spot stays round. Why?
3.)
I found a comment in Radiance Digest
The focal length produces only a subtle difference as it shifts the effective
position of the source behind the actual emitting surface.
I interpret this as shown in the lower image. Is this correct?
4.) Is the "specularity" parameter for the materials the same as "reflectivity" or "reflectance"?
1.) Each point in a ring is luminous when assigned a spotlight material, yes. However, you will only ever sample the center of the ring if you set "-dj 0" in your rendering, which is typically the default.
2.) This is an excellent question, and your drawing illustrates what goes on fairly accurately. The boundaries of the spotlight is dictated by the spotlight vector, but the output intensity is affected by the cosine of the ring orientation, since this changes the visible solid angle of the source.
3.) Your diagram looks right to me for the focal (spotlight vector) length. Effectively, it changes the fall-off of the source so that it is 1/r^2 measured from a point behind the object. The size of the spotlight area is unchanged -- i.e., the cone still covers the number of degrees specified.
Remember that a spotlight is just a convenient way to cut off the boundaries of illumination. It doesn't necessarily correspond to any real light source, but it can be a handy way to save calculation time for narrow spots, since there's no need to do shadow testing except for the part of the scene covered by the spotlight cone. Ideally, you would still use a measured light distribution or fall-off model as a pattern applied to your spotlight material.
4.) I haven't found a good definition of "reflectivity" that differs from "reflectance," but the specular component for a plastic is an uncolored highlight that corresponds to the given fraction of incident light. Generally speaking, non-metals do not have specularity values greater than 0.06 or so. The rest is treated as Lambertian, with the specified RGB color.
Does each point of e.g. a ring act as single spotlight when spotlight
is assigned to a surface? (upper left image)
2.)
When I assign a spotlight to a ring, how does the direction of the ring's normal
and the direction of the spotlight act together?
I would have expected the light spot to get elliptic when the spotlight is pointing
downwards but the ring has a different normal. (like the projected
area of the ring when looking in the direction of the light)
But the light spot stays round. Why?
3.)
I found a comment in Radiance Digest
The focal length produces only a subtle difference as it shifts the effective
position of the source behind the actual emitting surface.
I interpret this as shown in the lower image. Is this correct?
4.) Is the "specularity" parameter for the materials the same as "reflectivity" or "reflectance"?
thank you very much for your quick response and the good answers.
Abbout reflectivity / reflectance, I just found this
I am also using SolTrace, also from NREL, maybe you know it. It's a
more technical raytracer (path tracing, starting at the light source),
not considering colors, only energy.
Now I would like to make a rendering of a concentrating solar tower
plant. But since I have only very little experience with rendering so
far I can't get the materials right, or I am missing something else.
A solar tower plant consists of thousands of tracking mirrors reflection solar radiation
onto a receiver at the top of a tower.
But I was not able to see a light spot of the heliostat on the receiver (large black cylinder with lid).
I would have expected it at the spot marked in blue.
The scene is illuminated by a sun disk in zenith (0, 0, 1, from gendaylit) and a downward
pointing spot (red) directly above the heliostat (green). Heliostat diameter is 10 m.
This is the sun I am using at the moment
void light sun
0
3 7.131e+004, 7.131e+004, 7.131e+004
gendaylit originally returned
void light sun
0
3 7.131e+006, 7.131e+006, 7.131e+006
but this oversaturates my scene completely and it looks like a white-out.
I tried several materials for the heliostat (glossy colors, glass with no transmittance, mirror).
So far I tried a red plastic material for the receiver and played a bit with the specularity
and roughness but could not get it to reflect anything towards the camera.
Do you have some ideas how I could get this to work?
Kind regards,
Joe
···
Am 27.07.2017 19:37 schrieb Greg Ward:
Hi Joe,
I love your little drawings.
1.) Each point in a ring is luminous when assigned a spotlight
material, yes. However, you will only ever sample the center of the
ring if you set "-dj 0" in your rendering, which is typically the
default.
2.) This is an excellent question, and your drawing illustrates what
goes on fairly accurately. The boundaries of the spotlight is
dictated by the spotlight vector, but the output intensity is affected
by the cosine of the ring orientation, since this changes the visible
solid angle of the source.
3.) Your diagram looks right to me for the focal (spotlight vector)
length. Effectively, it changes the fall-off of the source so that it
is 1/r^2 measured from a point behind the object. The size of the
spotlight area is unchanged -- i.e., the cone still covers the number
of degrees specified.
Remember that a spotlight is just a convenient way to cut off the
boundaries of illumination. It doesn't necessarily correspond to any
real light source, but it can be a handy way to save calculation time
for narrow spots, since there's no need to do shadow testing except
for the part of the scene covered by the spotlight cone. Ideally, you
would still use a measured light distribution or fall-off model as a
pattern applied to your spotlight material.
4.) I haven't found a good definition of "reflectivity" that differs
from "reflectance," but the specular component for a plastic is an
uncolored highlight that corresponds to the given fraction of incident
light. Generally speaking, non-metals do not have specularity values
greater than 0.06 or so. The rest is treated as Lambertian, with the
specified RGB color.
Does each point of e.g. a ring act as single spotlight when spotlight
is assigned to a surface? (upper left image)
2.)
When I assign a spotlight to a ring, how does the direction of the ring's normal
and the direction of the spotlight act together?
I would have expected the light spot to get elliptic when the spotlight is pointing
downwards but the ring has a different normal. (like the projected
area of the ring when looking in the direction of the light)
But the light spot stays round. Why?
3.)
I found a comment in Radiance Digest
The focal length produces only a subtle difference as it shifts the effective
position of the source behind the actual emitting surface.
I interpret this as shown in the lower image. Is this correct?
4.) Is the "specularity" parameter for the materials the same as "reflectivity" or "reflectance"?
Regarding reflectivity, I suppose you could say that for any opaque material, reflectivity==reflectance. Since Radiance materials (with the possible exceptions of "dielectric," "interface" and "mist") don't really consider volumes, this equality generally applies.
I don't really know anything about SolTrace, but from your brief description, it sounds like it might be the right tool for your purpose.
The basic problem with Radiance is that as a light-backwards ray-tracer, it needs to know where to look for light sources, particularly tiny ones like the sun. There is a trick you can use, which is the "mirror" material light, which enables a virtual light source search. That brings other issue, especially once you go to a full array. By the way, it employs the "spotlight" mechanism internally to avoid unnecessary source tests, so you had a good idea with that.
I remember clearly that someone covered this in a Radiance workshop presentation. Can someone post the link for me? I haven't managed to locate it with any of my searches.
thank you very much for your quick response and the good answers.
Abbout reflectivity / reflectance, I just found this Reflectance - Wikipedia
I am also using SolTrace, also from NREL, maybe you know it. It's a
more technical raytracer (path tracing, starting at the light source),
not considering colors, only energy.
Now I would like to make a rendering of a concentrating solar tower
plant. But since I have only very little experience with rendering so
far I can't get the materials right, or I am missing something else.
A solar tower plant consists of thousands of tracking mirrors reflection solar radiation
onto a receiver at the top of a tower.
But I was not able to see a light spot of the heliostat on the receiver (large black cylinder with lid).
I would have expected it at the spot marked in blue.
The scene is illuminated by a sun disk in zenith (0, 0, 1, from gendaylit) and a downward
pointing spot (red) directly above the heliostat (green). Heliostat diameter is 10 m.
This is the sun I am using at the moment
void light sun
0
0
3 7.131e+004, 7.131e+004, 7.131e+004
gendaylit originally returned
void light sun
0
0
3 7.131e+006, 7.131e+006, 7.131e+006
but this oversaturates my scene completely and it looks like a white-out.
I tried several materials for the heliostat (glossy colors, glass with no transmittance, mirror).
So far I tried a red plastic material for the receiver and played a bit with the specularity
and roughness but could not get it to reflect anything towards the camera.
Do you have some ideas how I could get this to work?
I should add that Roland Schregle's photon mapping routines might help you with this calculation. Saves creating as many virtual light sources as you have mirror elements. (This number approaches infinity when you try to model curved mirrors.) Unfortunately, I am not knowledgeable enough about it to even tell you how to set it up, but I guess it will involve a portal for the sun, somewhere.
Cheers,
-Greg
···
From: Greg Ward <[email protected]>
Date: July 28, 2017 10:13:07 AM PDT
Hi Joe,
Regarding reflectivity, I suppose you could say that for any opaque material, reflectivity==reflectance. Since Radiance materials (with the possible exceptions of "dielectric," "interface" and "mist") don't really consider volumes, this equality generally applies.
I don't really know anything about SolTrace, but from your brief description, it sounds like it might be the right tool for your purpose.
The basic problem with Radiance is that as a light-backwards ray-tracer, it needs to know where to look for light sources, particularly tiny ones like the sun. There is a trick you can use, which is the "mirror" material light, which enables a virtual light source search. That brings other issue, especially once you go to a full array. By the way, it employs the "spotlight" mechanism internally to avoid unnecessary source tests, so you had a good idea with that.
I remember clearly that someone covered this in a Radiance workshop presentation. Can someone post the link for me? I haven't managed to locate it with any of my searches.
thank you very much for your quick response and the good answers.
Abbout reflectivity / reflectance, I just found this Reflectance - Wikipedia
I am also using SolTrace, also from NREL, maybe you know it. It's a
more technical raytracer (path tracing, starting at the light source),
not considering colors, only energy.
Now I would like to make a rendering of a concentrating solar tower
plant. But since I have only very little experience with rendering so
far I can't get the materials right, or I am missing something else.
A solar tower plant consists of thousands of tracking mirrors reflection solar radiation
onto a receiver at the top of a tower.
But I was not able to see a light spot of the heliostat on the receiver (large black cylinder with lid).
I would have expected it at the spot marked in blue.
The scene is illuminated by a sun disk in zenith (0, 0, 1, from gendaylit) and a downward
pointing spot (red) directly above the heliostat (green). Heliostat diameter is 10 m.
This is the sun I am using at the moment
void light sun
0
0
3 7.131e+004, 7.131e+004, 7.131e+004
gendaylit originally returned
void light sun
0
0
3 7.131e+006, 7.131e+006, 7.131e+006
but this oversaturates my scene completely and it looks like a white-out.
I tried several materials for the heliostat (glossy colors, glass with no transmittance, mirror).
So far I tried a red plastic material for the receiver and played a bit with the specularity
and roughness but could not get it to reflect anything towards the camera.
Do you have some ideas how I could get this to work?
He was using Radiance like SolTrace, to calculate the flux distribution on the receiver.
He calculated the surface normals at the receiver panels sent them
into the scene using rtrace.
I am not sure if I understood that correctly:
The material "mirror" uses sort of mimics the spotlight, so only
shooting rays within a fixed cone angle towards the light sources
or other illumination surfaces?
I will also look into the photon mapping, have never used that before.
Well, I would like to create realistic renderings of these power plands at different values of solar irradiance, e.g. 500 and 1000 W / m^2.
But instead of "cheating" myself into a glowing receiver, I would
rather have it to be illuminated by the heliostats.
Maybe also with some mist:
Do you think it is possible to create images like that at all?
Kind regards and thanks for your patience
Joe
···
Am 28.07.2017 um 19:24 schrieb Greg Ward:
I should add that Roland Schregle's photon mapping routines might help you with this calculation. Saves creating as many virtual light sources as you have mirror elements. (This number approaches infinity when you try to model curved mirrors.) Unfortunately, I am not knowledgeable enough about it to even tell you how to set it up, but I guess it will involve a portal for the sun, somewhere.
Cheers,
-Greg
From: Greg Ward <[email protected]>
Date: July 28, 2017 10:13:07 AM PDT
Hi Joe,
Regarding reflectivity, I suppose you could say that for any opaque material, reflectivity==reflectance. Since Radiance materials (with the possible exceptions of "dielectric," "interface" and "mist") don't really consider volumes, this equality generally applies.
I don't really know anything about SolTrace, but from your brief description, it sounds like it might be the right tool for your purpose.
The basic problem with Radiance is that as a light-backwards ray-tracer, it needs to know where to look for light sources, particularly tiny ones like the sun. There is a trick you can use, which is the "mirror" material light, which enables a virtual light source search. That brings other issue, especially once you go to a full array. By the way, it employs the "spotlight" mechanism internally to avoid unnecessary source tests, so you had a good idea with that.
I remember clearly that someone covered this in a Radiance workshop presentation. Can someone post the link for me? I haven't managed to locate it with any of my searches.
thank you very much for your quick response and the good answers.
Abbout reflectivity / reflectance, I just found this Reflectance - Wikipedia
I am also using SolTrace, also from NREL, maybe you know it. It's a
more technical raytracer (path tracing, starting at the light source),
not considering colors, only energy.
Now I would like to make a rendering of a concentrating solar tower
plant. But since I have only very little experience with rendering so
far I can't get the materials right, or I am missing something else.
A solar tower plant consists of thousands of tracking mirrors reflection solar radiation
onto a receiver at the top of a tower.
But I was not able to see a light spot of the heliostat on the receiver (large black cylinder with lid).
I would have expected it at the spot marked in blue.
The scene is illuminated by a sun disk in zenith (0, 0, 1, from gendaylit) and a downward
pointing spot (red) directly above the heliostat (green). Heliostat diameter is 10 m.
This is the sun I am using at the moment
void light sun
0
3 7.131e+004, 7.131e+004, 7.131e+004
gendaylit originally returned
void light sun
0
3 7.131e+006, 7.131e+006, 7.131e+006
but this oversaturates my scene completely and it looks like a white-out.
I tried several materials for the heliostat (glossy colors, glass with no transmittance, mirror).
So far I tried a red plastic material for the receiver and played a bit with the specularity
and roughness but could not get it to reflect anything towards the camera.
Do you have some ideas how I could get this to work?
The only time specularity == reflectance is when you have a plastic material whose first three (RGB) arguments are all zero. Otherwise, there is a diffuse component that gets added to the specular in the following manner:
Good -- you found the presentation from Germain. That's the one I was looking for.
Your understanding of the "mirror" material is correct. For every light source in the scene, a virtual spotlight is created that encompasses the surface modified by "mirror", "prism", or "prism2". The number of reflections searched is set by the "-dr" option, which should be set to 1 in this case because you neither care about nor wish to burden the calculation searching for secondary reflections.
You should be able to create images using the virtual light source calculation Germain used. The photon-mapping approach ought to work as well. You don't need to "cheat," although getting a reasonable tone-mapped exposure might be challenging. You can play around with pcond, especially the veiling glare simulation, to give you the kind of flare effect you are looking for in a post-process.
He was using Radiance like SolTrace, to calculate the flux distribution on the receiver.
He calculated the surface normals at the receiver panels sent them
into the scene using rtrace.
I am not sure if I understood that correctly:
The material "mirror" uses sort of mimics the spotlight, so only
shooting rays within a fixed cone angle towards the light sources
or other illumination surfaces?
I will also look into the photon mapping, have never used that before.
Well, I would like to create realistic renderings of these power plands at different values of solar irradiance, e.g. 500 and 1000 W / m^2.
Do you think it is possible to create images like that at all?
Kind regards and thanks for your patience
Joe
Am 28.07.2017 um 19:24 schrieb Greg Ward:
I should add that Roland Schregle's photon mapping routines might help you with this calculation. Saves creating as many virtual light sources as you have mirror elements. (This number approaches infinity when you try to model curved mirrors.) Unfortunately, I am not knowledgeable enough about it to even tell you how to set it up, but I guess it will involve a portal for the sun, somewhere.
Cheers,
-Greg
From: Greg Ward <[email protected]>
Date: July 28, 2017 10:13:07 AM PDT
Hi Joe,
Regarding reflectivity, I suppose you could say that for any opaque material, reflectivity==reflectance. Since Radiance materials (with the possible exceptions of "dielectric," "interface" and "mist") don't really consider volumes, this equality generally applies.
I don't really know anything about SolTrace, but from your brief description, it sounds like it might be the right tool for your purpose.
The basic problem with Radiance is that as a light-backwards ray-tracer, it needs to know where to look for light sources, particularly tiny ones like the sun. There is a trick you can use, which is the "mirror" material light, which enables a virtual light source search. That brings other issue, especially once you go to a full array. By the way, it employs the "spotlight" mechanism internally to avoid unnecessary source tests, so you had a good idea with that.
I remember clearly that someone covered this in a Radiance workshop presentation. Can someone post the link for me? I haven't managed to locate it with any of my searches.
thank you very much for your quick response and the good answers.
Abbout reflectivity / reflectance, I just found this Reflectance - Wikipedia
I am also using SolTrace, also from NREL, maybe you know it. It's a
more technical raytracer (path tracing, starting at the light source),
not considering colors, only energy.
Now I would like to make a rendering of a concentrating solar tower
plant. But since I have only very little experience with rendering so
far I can't get the materials right, or I am missing something else.
A solar tower plant consists of thousands of tracking mirrors reflection solar radiation
onto a receiver at the top of a tower.
But I was not able to see a light spot of the heliostat on the receiver (large black cylinder with lid).
I would have expected it at the spot marked in blue.
The scene is illuminated by a sun disk in zenith (0, 0, 1, from gendaylit) and a downward
pointing spot (red) directly above the heliostat (green). Heliostat diameter is 10 m.
This is the sun I am using at the moment
void light sun
0
0
3 7.131e+004, 7.131e+004, 7.131e+004
gendaylit originally returned
void light sun
0
0
3 7.131e+006, 7.131e+006, 7.131e+006
but this oversaturates my scene completely and it looks like a white-out.
I tried several materials for the heliostat (glossy colors, glass with no transmittance, mirror).
So far I tried a red plastic material for the receiver and played a bit with the specularity
and roughness but could not get it to reflect anything towards the camera.
Do you have some ideas how I could get this to work?