Improved Lighting Model in Alto Clouds causing black specs

[shadowphile]

Specifically the specs are as in this pic.  They go away if I uncheck Improved Lighting Model.  My scene is not using GI lighting either.
Tried lots of other quality tweaks, this is the only one that works.
thanks
[attachimg=1]

[Tangled-Universe]

Can you share the .tgd file? That would be very useful.

Cheers,
Martin

[gsmith]

Which version of TG2 are you using?

[shadowphile]

I am using the latest, 2.3.20.1, 64-bit version
Win 7 Pro OS. 4G RAM

Attached is the the file stripped to the bare minimum.  Just open and hit render.

thanks

[Tangled-Universe]

Thanks, I will test it and report it to the bug-tracker.

Cheers,
Martin

[FrankB]

there are no black spots when I render this file. But it seems the file you sent has less cloud layers than the one you posted. I suggest you post a version with the whole atmosphere part unaltered.

Frank

[Tangled-Universe]

Hmmm ok...his post suggests that the file he submitted also shows the problem, but apparently it isn't.
I haven't had the chance to check it yet, but I think he should post the un-altered version anyway.

[Tangled-Universe]

Ok just tested this and it also renders black spots...

edit:

Figured it out: it's not the improved glow model which causes the black spots, but it is the negative setting for "fake internal scattering".
Set that to 0 or >0 and the black spots will disappear.

I'll explain you briefly what fake internal scattering does:

Light propagation is the amount of light passing through the cloud.
Light propagation mix is the weight of how the scattering is being calculated.
A value of 0.5 is 50% scattering by light propagation and 50% by fake internal scatter.
Fake internal scatter causes thin clouds to be brighter at the edges (mostly), because of more forward scatter.
With dense clouds there will be more back scatter and thus darker edges.
A fake internal scatter of 1 equally distributes forward and backward scattering.

Recently anisotropic scattering is being added to clouds and that had great impact on how these 3 values interact.

Here's a quote from a change-log on anisotropic scattering and fake internal scatter:

- New option "anisotropic enviro light". Enabled by default. When enabled, enviro light (GI or Ambient Occlusion) contributions are weighted more strongly in the forward viewing direction than the backward viewing direction for thin clouds, and more strongly in the backward direction than the forward direction for dense clouds. The "fake internal scattering" parameter controls the change from forward scattering in thin cloud to back scattering in dense clouds, as well as brightening the contribution of direct light sources on dense clouds. If fake internal scattering is 0 while anisotropic enviro light is enabled, cloud density has no effect on enviro light contributions and they are predominantly forward scattering everywhere. The "sun glow" settings also affect the contrast between forward- and back-scattering weightings.

So I think the change from -0.5 to 0 for fake internal scattering won't affect your work, other than fixing the issue.

Cheers,
Martin

[FrankB]

hmm.. so why have there been no black spots when I rendered this? Didn't change anything, just hit render.

[Tangled-Universe]

hmm.. so why have there been no black spots when I rendered this? Didn't change anything, just hit render.

Yeah that's puzzling, to be honest...

Did you render it with the public version?

[jaf]

I got the black spec using the x64 version and the x32 version.  But it's kind of weird -- the black specs don't match.  I just loaded each x32/x64 version and did a render. The file was set to do a crop render.

[Tangled-Universe]

The black spots don't need to match.

I tested the file for solutions and thus made >dozen renders. Each time the pattern was different.

Since it probably has to do with the scatter settings (see my previous post) which is influenced by enviro light which in turn is GI related, I think it has to do with the known fact that each (crop) render has slightly different GI solution/calculation.
Therefore it wouldn't surprise that the pattern of black spots is different with each render or across 32/64 bit.