toilet - something's wrong...

[wojtek-w]

hi there,
i'm fresh in indigo and still learning... My main problem is rendering time. The pictures attached below is a result of 11 hours of computing - with resolution 800x800 is much too long (imho). As you can see most of unwanted noise became from leyer 1 (ceiling mounted lights made with emitting meshes). Layer 0 is exit portal. Of course i read and followed some basic tips found on forum (using "exit portal" shading, no 100 percent whiteness (max 240/240/240 in rgb), MRI lowered to 200). Any ideas how to improve model, scene or settings to make it render faster?

I use sketchup 7 with skindigo.
CPU is intel core 2 duo 3 Ghz, Gforce 8600 gt, 2Gb RAM
thanks for any suggestion...

[Zom-B]

Your "Layer 0" seems quite noisefree...

The reason is that your second layer uses a way lower light flux than layer0. Indigo dives rendering power among the layers by this light flux. If your layer0 has 100times more light flux, than it gets 100 times more samples calculated then the other layer!!!

Try using linear or camera tonemapping to see this issue. The used Flux is shown after starting Indigo at the end of Scene calculation. Ajust the Flux by using emission Scale. if a lightsource gets to bright use the layer settings to reduce it.

I hope this is understandable...



Another tweak to raise render times is having your lightsources simple diffuse material with absolute black as diffuse color. By this the emitter mesh don't reflect any light and this speed up theother ray reflections etc.

...no 100 percent whiteness (max 240/240/240 in rgb), MRI lowered to 200).

I don't know what MRI means, but just to make it clear: A max 80% of saturation (for each color!) is the way to go (max 204 on RGB)...

[wojtek-w]

thank you very much.
i will follow your advises.
i'm just not sure if i understood - the problem is a difference in light power between layers, right?
i don't know how can I reduce the power of exit portal - there's no emitter parametter in material editor when you use exit portal.
I can always increase power of emitting meshes, but in real life i wouldn't use f.e. 5x500w halogen in such a small space.

Try using linear or camera tonemapping to see this issue. The used Flux is shown after starting Indigo at the end of Scene calculation. Ajust the Flux by using emission Scale. if a lightsource gets to bright use the layer settings to reduce it.

changing any parametter in linear or reinhardt tonemapping doesn't affect camera tonemapping, so if i want to use camera there's no point to chang linear scale, right? I looked at render log and didn't found anything refering to flux.

I don't know what MRI means

I messed up i meant MNCR

A max 80% of saturation (for each color!) is the way to go (max 204 on RGB)

And what about map using materials? Should I desaturate jpgs too?
Thank again

[WytRaven]

Your "Layer 0" seems quite noisefree...

The reason is that your second layer uses a way lower light flux than layer0. Indigo dives rendering power among the layers by this light flux. If your layer0 has 100times more light flux, than it gets 100 times more samples calculated then the other layer!!!

So does this not then almost automatically beg for a feature that allows us to mark a layer as "complete" so indigo stops working on it and focuses all power on the remaining layers? Now that to me would really make light layers an awesome feature as you get flexibility and a serious human controlled optimistation. Unless of course I am completely misunderstanding your post ZomB...

[Zom-B]

i'm just not sure if i understood - the problem is a difference in light power between layers, right?

jep!

i don't know how can I reduce the power of exit portal - there's no emitter parametter in material editor when you use exit portal.
I can always increase power of emitting meshes, but in real life i wouldn't use f.e. 5x500w halogen in such a small space.

A Exit Portal is only a opening for your Sun/sky or EnvMap light. Adjusting EnvMaps is possible, since they are based on a material that you create and tweak.
A workaround would be not to use layers, or simply as I told raise the power of your lamp x100 and then set the overall brightness of the layer to 0,01. This system isn't realy operfect at the moment, but well...

changing any parametter in linear or reinhardt tonemapping doesn't affect camera tonemapping, so if i want to use camera there's no point to chang linear scale, right?

each tonemapping algo is independent from each other in the settings... I adviced you to try cam or linear just to realize how dark your second light layer is. Reinhard is adaptive tonemapper that checkes the image and reveals everything.

I looked at render log and didn't found anything refering to flux.

Here a a snipped out of the "Indigo_Competition_Entry_-_Erotica-FINAL-Oleg_Bogattke_aka_psor.igs", the value is at the top here:

Code: Select all

Light luminous flux (geometry name=Studio_Light.001): 9.600E+004 lm
Building Object Tree...
	6 objects.
	calcing root AABB.
	AABB: (-0.773170, -2.251853, -0.000000), (3.104088, 1.684535, 3.669422)
	max tree depth: 7
	reserving N nodes: 6(48 B)
	total nodes used: 11 (88 B)
	total leafgeom size: 11 (88 B)
Finished building tree.
AutoFocus: setting camera focus distance to 0.73851 m.
Num buffer layers: 1
Master buffer size: 30.473 MB
Auxiliary buffer size: 100.893 MB
Creating diffraction filter image...
	Done.  (Elapsed: 3.08329 s)
Settings:
	Image width: 920 px
	Image height: 720 px
	Internal image width: 1844 px
	Internal image height: 1444 px
	Display period: 120.00000 s
	Image save period: 60.00000 s
	Frame upload period: 40.00000 s
	Splat filter: mn_cubic, blur=0.333000, ring=0.333000
	Downsize filter: mn_cubic, blur=0.33300, ring=0.33300
	Render region: false
	Supersample factor: 2
	Metropolis: true
	Bidirectional: true
	Hybrid: false
	Aperture diffraction: true
	Post-process diffraction: true
Auto setting number of threads to 2.
Finished initialisation (Time Taken: 14.65957 s)

And what about map using materials? Should I desaturate jpgs too?

This should be possible by tweaking the b parameter of the texture... I'm not perfectly sure here at the moment

So does this not then almost automatically beg for a feature that allows us to mark a layer as "complete" so indigo stops working on it and focuses all power on the remaining layers?

yes, this really begs for such a feature
Ono once told be about this automated control of priority for light Layers in IRC, since I reported it as a Bug ^^

[CTZn]

And what about map using materials? Should I desaturate jpgs too?

This should be possible by tweaking the b parameter of the texture... I'm not perfectly sure here at the moment

I said that but this is not correct, playing with a, b and c will only affect the value (luminosity), not the saturation.

It may be possible to process the texture through an ISL pipeline to tweak its saturation, there was something close done by msuDom IIRC. But for now yes, avoid saturated textures.