VRayLight directionality

[Art48]

Hi,

Any forumla around which describes the connection of directionality and intensity int he Plane Light?
What would be really awesome is if there was a checkbox where we could enable a feature to keep the intensity of a light disconncted from the directionality. Kind of how the spotLight works of 3ds max standard.
Untril then I would like to know some formula where I can script some reverse controller for the intensity connected to the directionality

[voltron7]

see wishlist, I think it's there many times...
Link

Guess it's on the far back burner

[Vizioen]

Hi,

Any forumla around which describes the connection of directionality and intensity int he Plane Light?
What would be really awesome is if there was a checkbox where we could enable a feature to keep the intensity of a light disconncted from the directionality. Kind of how the spotLight works of 3ds max standard.
Untril then I would like to know some formula where I can script some reverse controller for the intensity connected to the directionality

I didn't even know this option existed. But I can see the benefit of it not affecting the multiplier, so +1 from me

[Art48]

FOr the time being a factor or some kind of formula would be cool so we can script some tools ourselves

[Vizioen]

Can't this be measured?

[Art48]

I guess it could but its kind of time intensive and prone to error due to color mapping colo etc and time investment.
I'm sure one of that hardcore math devs can come up with a formula, maybe they even have it already I mean somehow it has to be calculated anyways :P

[Vizioen]

I guess it could but its kind of time intensive and prone to error due to color mapping colo etc and time investment.
I'm sure one of that hardcore math devs can come up with a formula, maybe they even have it already I mean somehow it has to be calculated anyways :P

Well I tried, I don't have the patience/time to analyze the data anymore. But here are my findings so maybe you can figure this out.

I took 10 images with a single plane light. No GI. No color mapping. And measured the diffuse in the same point. Vraycamera at 200 Shutter and 100 ISO. Gamma 2.2

You can find all the files in the following zip: http://public.vizioen.be/chaos/lighttest.zip

Directional: 0

Directional: 0.1

Directional: 0.2

Directional: 0.3

Directional: 0.4

[Vizioen]

Directional: 0.5

Directional: 0.6

Directional: 0.7

Directional: 0.8

Directional: 0.9

[Vizioen]

Directional: 1


This is an excel file of the values


And this is a graph made in Rhino of the values. As you can see it's pretty linear. But there's an inconsistency in the beginning it dips down first and then goes up in a linear fashion. What this graph doesnt show is how fast it rises though.



edit: this bar shows how fast it rises...weird

[Vizioen]

I've done another test now. I divided the numbers of 0 and 0.5 and 0.5 was about 4 times stronger (4,115898058 to be exact) in values in that specific point on the floor. So I've divided my multiplier of 400 by 4 and the result on the floor is pretty consistent with the other one I think.

[Art48]

Thanks! COool stuff... So it would be inverse square? I dunno I am no mathematician :P.
If I always half the light intensity when doubling the directionality the formula would be? Can't wrap my head around it :P

[theedge]

Sadly not good in math neither. But here's my 2 cents :

The light is emitting the same intensity (I) whatever the directional parameter is set to.
So the more the light beam is spread, the larger is the lit area (A) and the smaller is the "density" of light (dL) : dL=I/A

So we want to obtain not the same light intensity but the same light "density" on the lit surface.

There must be some relationship between the spread angle and the lit surface increase. And because we have an angle this relationship must involve a bit of trig.

[Svetlozar Draganov]

Sadly not good in math neither. But here's my 2 cents :

The light is emitting the same intensity (I) whatever the directional parameter is set to.
So the more the light beam is spread, the larger is the lit area (A) and the smaller is the "density" of light (dL) : dI=I/A

So we want to obtain not the same light intensity but the same light "density" on the lit surface.

There must be some relationship between the spread angle and the lit surface increase. And because we have an angle this relationship must involve a bit of trig.

I've just discussed this with one of our developers but unfortunately he informed me that there isn't an easy formula to calculate the exact intensity value in order to preserve the same illumination between different values of Directional parameter. It seams the things are a bit more complicated than they appears.

[Art48]

Okay that's good to know. I mean basically an approximation formula would be enough. Just that it roughly keepos the same visible intensity

[rajdarge]

So I have been puzzling this thing out for so long now. Its pretty important as soon as you make a light directional, the exposure gets blown out (often).
This is a script that should be loaded into the editor in a blank scene . It will generate 91 point objects that you can delete with delete $point*
Here is the code:

animate off
LtTesttarg = targetobject pos:[0,0,0] name:"TestLight.target"
LtTest = vraylight pos:[0,0,100] targeted:true name:"TestLight"  width:5 height:5 normalizecolor:3 multiplier:1.0 preview_distribution:1 lightDistribution:0.0
LtTest.target = LtTesttarg
ltMeter = VRayLightMeter active:true width:10 length:10 orient:true colors:true text:true max:100000.0 min:0.0 pos:[0,0,0]
MeterReadings = #()
GraphPoints = #()
with animate on 
(
	at time 0 ltTest.lightDistribution = 0.0 
		
	at time 90 ltTest.lightDistribution = 1.0
		
		
)	
ltTest.lightDistribution.controller.keys[1].outTangentType = #linear
ltTest.lightDistribution.controller.keys[2].inTangentType = #linear
for i = 0 to 90 do 
(
	slidertime = i
	ltMeter.calculate()
	MeterReadings[i+1] = ltMeter.direct[5]
	format "Angle: % Reading: % \n" i MeterReadings[i+1]
	if (MeterReadings[1] != 0 or MeterReadings[1] != undefined)do	GraphPoints[i+1] = point pos:[i,-20,MeterReadings[i+1]/MeterReadings[1]] cross:false centermarker:true
)

The output is proportional output dependent on when the directional value is 0
it looks like a exponential curve with a sine wave superimposed and a plateau at around 87 degrees
can someone translate that into an equation?
So are there any mathemeticians out there?
Raj