Intensity of Sun and Sky

Discussion related to the implementation of new features & algorithms to the Core Engine.

Moderators: Dade, jromang, tomb, zcott, coordinators

Re: Intensity of Sun and Sky

Postby Dade » Sun Feb 19, 2017 7:10 am

HoloFan wrote:Why would there be another 1/pi? Is that for compensating some spherical integral I am not aware of?


Irradiance is the result of an integration, if you check PBRT book (second edition) pag. 284 irradiance is:

E = d Flux / d Area (differential flux over differential area)

and in our case dA is PI (the area of a circle of radius 1.0).

P.S. LuxCore resembles many aspects of PBRT (because is derived from classic Lux that, in turn, was derived from PBRT v1) but there is also some major difference. Try to do rendering of the same scene with LuxCore and PBRT v2, LuxCore is about 10-100 times faster. It is something I have seen during some test I have done in the past. I was a bit surprised, the difference is HUGE (and I'm talking of CPUs, I'm not even starting to put GPUs in the mix).
User avatar
Dade
Developer
 
Posts: 8318
Joined: Sat Apr 19, 2008 6:04 pm
Location: Italy

Re: Intensity of Sun and Sky

Postby crosley09 » Mon Feb 20, 2017 12:36 pm

Dade wrote:Just to be clear, Lux doesn't know anything outside the visible spectrum. For instance, if you have an infra-red light source of 100watt, you are supposed to have an efficacy of 0 (nothing is visible) and you will obtain an irradiance value of 0 Watt/m^2 because Lux doesn't transport anything out side the visible spectrum.

Inside the visible spectrum however watt/m^2 to lux is just a unit transformation (like the one done by the contour lines plugin).


How can you work in watt/m^2 then? When Watts * Efficacy = Lumens, the output of the "irradiance AOV" is in lumens because efficacy is taken into account.

This may just be a matter of using the right units, but it seems that the Irradiance AOV should be call the Illuminance AOV with units lumens and lux instead watts and watts/m^2. Otherwise, efficacy should be removed from the equation and factored in elsewhere.

I hope this makes some sense, and it is entirely possible that i'm misunderstanding again.
i7 5930k, GTX 980 ti, 32 GB ddr4, 512 GB PCIe SSD...Windows 10
User avatar
crosley09
 
Posts: 199
Joined: Fri Oct 07, 2011 1:00 pm
Location: Indiana, USA

Re: Intensity of Sun and Sky

Postby HoloFan » Sat Feb 25, 2017 1:48 pm

Dade wrote:Irradiance is the result of an integration, if you check PBRT book (second edition) pag. 284 irradiance is:

E = d Flux / d Area (differential flux over differential area)

and in our case dA is PI (the area of a circle of radius 1.0).

Irradiance can be the result of an integration (as on page 288). But you could also integrate the Irradiance over an area A and you get the flux of energy through the surface you integrated over.

Maybe we are talking about different situations. I thought we were talking about the irradienance on any point of a unit sphere with a 100W light source at its center. For that I would use the first formula on p.284: "Irradiance at any point inside A_1 is then E_1 = Phi/A_1" (p.284).
In that case A would be the area of the sphere (4*pi 1^2). Why would we need the area of a circle? Or did we just discuss different questions? Anyway it's not that important to discuss it over ten posts or something. I would much rather see new cool features than taking up your time with such stuff. ;)

Dade wrote:Try to do rendering of the same scene with LuxCore and PBRT v2, LuxCore is about 10-100 times faster.


That's really impressive. But I can believe it from experience. When I began it was like "Start the render, go for a half hour walk, then look whether you can see enough." Now it's more like: "Sadly it takes a few seconds and not milliseconds to resolve." I just never directly compared them on the same hardware. :)
Thank you very much for your efforts to this great render engine!
HoloFan
 
Posts: 103
Joined: Sat May 14, 2011 7:04 am

Re: Intensity of Sun and Sky

Postby sharlybg » Fri Mar 31, 2017 8:40 am

I've read all the post here and finally i'm confused. Why ?

I'm currently building a specific workflow with luxcore and blender. And it is intended to use realworld value and parameter so i use realworld camera set and realworld light. I'm currently comparing luxcore to Vray 3.4 (camera/light). This is results we got.

Camera setting for Vray:

fnumber = 8
ISO = 100
Shutter = 50 ( i think this mean 1/50)

Image


Camera setting for Luxcore:

fnumber = 8
ISO = 100
Shutter = 1/50

Image


Vray and Luxcore both use Preetham sky model.
For Lux camera i've use SHUTTER NORMALISE (i feel that absolute give to much brighter result).

Here is the scene if you want give it a try.

Camset.blend
(478.9 KiB) Downloaded 11 times
i7 6700k + 32 Gb DDR4 + 2X R9 390 sapphir nitro.
User avatar
sharlybg
 
Posts: 687
Joined: Tue Nov 02, 2010 10:22 am
Location: Ivory coast

Re: Intensity of Sun and Sky

Postby Dade » Fri Mar 31, 2017 12:26 pm

sharlybg wrote:Vray and Luxcore both use Preetham sky model.


You seem to render a scene with a Sun+Sky light sources, have you tried to render a scene with sky only ?
User avatar
Dade
Developer
 
Posts: 8318
Joined: Sat Apr 19, 2008 6:04 pm
Location: Italy

Re: Intensity of Sun and Sky

Postby sharlybg » Sat Apr 01, 2017 3:49 am

You seem to render a scene with a Sun+Sky light sources, have you tried to render a scene with sky only ?


Yes this is what we are testing so far ! Or am i wrong ?

if i want to test daylight i need sunsky model in luxrender.
i7 6700k + 32 Gb DDR4 + 2X R9 390 sapphir nitro.
User avatar
sharlybg
 
Posts: 687
Joined: Tue Nov 02, 2010 10:22 am
Location: Ivory coast

Re: Intensity of Sun and Sky

Postby Dade » Sat Apr 01, 2017 4:07 am

sharlybg wrote:
You seem to render a scene with a Sun+Sky light sources, have you tried to render a scene with sky only ?


Yes this is what we are testing so far ! Or am i wrong ?


You are, there is a sun in the above renders: otherwise, you could not have sharp illumination like the the window on the floor. Sun and Sky are placed together as an utility by LuxBlend but you can pick sky or sun alone too (and I assume VRay is doing the same).
User avatar
Dade
Developer
 
Posts: 8318
Joined: Sat Apr 19, 2008 6:04 pm
Location: Italy

Re: Intensity of Sun and Sky

Postby Dade » Sat Apr 01, 2017 4:13 am

Btw, LuxRender uses Hosek's sky model: http://cgg.mff.cuni.cz/projects/SkylightModelling/

Preetham's sky model is quite old as far as I remember and has been surpassed by Hosek's model (more information available in the linked paper: http://cgg.mff.cuni.cz/projects/Skyligh ... eprint.pdf)
User avatar
Dade
Developer
 
Posts: 8318
Joined: Sat Apr 19, 2008 6:04 pm
Location: Italy

Re: Intensity of Sun and Sky

Postby B.Y.O.B. » Sat Apr 01, 2017 4:21 am

sharlybg wrote:For Lux camera i've use SHUTTER NORMALISE (i feel that absolute give to much brighter result).

Um wat :?
Didn't we have this recently? viewtopic.php?f=16&t=13312#p126169
Code: Select all
    if self.exposure_mode == 'normalised':
            time = (self.exposure_end_norm - self.exposure_start_norm) / fps

So your shutter time will be (1/50) / 25 if you set 25 fps, which is 0.0008 s instead of the 0.02 s it should be.
So set it to absolute if you really want 1/50 s shuttertime.
User avatar
B.Y.O.B.
Developer
 
Posts: 5095
Joined: Wed Nov 10, 2010 4:10 pm
Location: Germany

Re: Intensity of Sun and Sky

Postby sharlybg » Sat Apr 01, 2017 6:52 am

Dade wrote:Btw, LuxRender uses Hosek's sky model: http://cgg.mff.cuni.cz/projects/SkylightModelling/

Preetham's sky model is quite old as far as I remember and has been surpassed by Hosek's model (more information available in the linked paper: http://cgg.mff.cuni.cz/projects/Skyligh ... eprint.pdf)


Here is the result with sky only hosek's.LUXCORE

fnumber = 8
ISO = 100
Shutter = 50 (absolute mode)

camtest.jpg



So If i want to mimic daylight in luxcore i've to use normal SUNSKY . If yes, is the implementation inside Luxcore is correct enought to use realworld value as reference ?

B.Y.O.B. wrote:
sharlybg wrote:For Lux camera i've use SHUTTER NORMALISE (i feel that absolute give to much brighter result).

Um wat :?
Didn't we have this recently? viewtopic.php?f=16&t=13312#p126169
Code: Select all
    if self.exposure_mode == 'normalised':
            time = (self.exposure_end_norm - self.exposure_start_norm) / fps

So your shutter time will be (1/50) / 25 if you set 25 fps, which is 0.0008 s instead of the 0.02 s it should be.
So set it to absolute if you really want 1/50 s shuttertime.


Don't worry my friends I'm very very confused now.

1/ For the render above it's another test i've done with a friend with vray and 3dsmax licence

2/ in this test when i use my Shutter speed in absolute mode i get far brighter render than Him so decied to switch to normalised (but i've forget that this setting is not for still render)

So is there a possibility to mimic realworld camera/natural light/ artificial light inside luxcore ?
i7 6700k + 32 Gb DDR4 + 2X R9 390 sapphir nitro.
User avatar
sharlybg
 
Posts: 687
Joined: Tue Nov 02, 2010 10:22 am
Location: Ivory coast

PreviousNext

Return to Architecture & Design

Who is online

Users browsing this forum: No registered users and 1 guest