'Building scene' - Indigo freezes

[Meelis]

Closed box 2 x 2 x 2 m
2 emitters
one plane 1 x 0.12 m, and one 96 faces spher 1 cm or 0.01 m in diameter
with material diffuse - color 0.8 0.8 0.8 - Luminous Flux 3000 / gain 1 / RGB 1 1 1

OnoSendai wrote here http://www.indigorenderer.com/forum/vie ... f=7&t=6567

1) Use uniform scattering coefficient
2) Make sure the absorption coefficient isn't too low compared to the scattering coefficient, otherwise light will scatter around for ages.
3) Make sure the scattering coefficient isn't too high, for the same reason.

Yeah with uniform scattering coefficient it renders or if RGB values are more or less under 10 (or not 100, below that).

That material works even with uniform value 10000 (yes slow with 10000)

Code: Select all

<?xml version="1.0" encoding="utf-8"?>

<scenedata>
	<medium>
		<name>Milk</name>
		<precedence>5</precedence>
		<basic>
			<ior>1.34</ior>
			<cauchy_b_coeff>0</cauchy_b_coeff>
			<absorption_coefficient_spectrum>
				<rgb>
					<rgb>8.689 8.689 8.689</rgb>
					<gamma>2.2</gamma>
				</rgb>
			</absorption_coefficient_spectrum>
			<subsurface_scattering>
				<scattering_coefficient_spectrum>
					<uniform>
						<value>1000</value>
					</uniform>
				</scattering_coefficient_spectrum>
				<phase_function>
					<henyey_greenstein>
						<g_spectrum>
							<rgb>
								<rgb>0.75 0.714 0.681</rgb>
								<gamma>2.2</gamma>
							</rgb>
						</g_spectrum>
					</henyey_greenstein>
				</phase_function>
			</subsurface_scattering>
		</basic>
	</medium>
	<material>
		<name>Milk</name>
		<specular>
			<transparent>true</transparent>
			<internal_medium_name>Milk</internal_medium_name>
		</specular>
	</material>

</scenedata>

[CTZn]

The problematic material works fine in my case if there is a light plane behind, but doesn't start if sun is on. Make sure that you are not using the sun if you are shooting in a box. Also emitters are best of black albedo.

[Meelis]

I got the material to render - posted it under the name "What's wrong with that material?"

I set the whole box to black and all walls as emitters and building scene took 11 min.
But it's much slower than uniform 10 000


Click for full size

[Zom-B]

welcome to the next level of SSS

RGB based SSS is in general way slower then uniform based!
Raising the "absorption_coefficient_spectrum" is needed to get some acceptable speed out of this material...
And by raising I mean a uniform multi to RGB by 10 or 100!

I basically use always uniform SSS and color it by tweaking the Absorption layer, for my few test I did this way the result so far.

[Meelis]

welcome to the next level of SSS

RGB based SSS is in general way slower then uniform based!
Raising the "absorption_coefficient_spectrum" is needed to get some acceptable speed out of this material...
And by raising I mean a uniform multi to RGB by 10 or 100!

I basically use always uniform SSS and color it by tweaking the Absorption layer, for my few test I did this way the result so far.

Yeah insane speed 54 s/sec on home 2 computer network.

But nice thing now i know by using more isotropic H. G phase function, it will give medium the look like absorption
is higher and i don't have to use so high sss uniform value to get reall looking material.

[Zom-B]

The low speed is caused by the big scale difference in absorption_coefficient_spectrum to the scattering_coefficient_spectrum.

Also for uniform SSS you can enhance speed by narrowing the absorption_coefficient_spectrum to the scattering!