Broken reflections (chunk-wise)

[Volker Harun]

Hi there,

I am trying to get the acquainted with my ancient WIPs.
Attached is a render with a reflective surface. It is displaced with a soft PF to an amount of 0.05. This should not be the issue. A cropped render without this displacement is looking the same. Each rendered chunk reflects different.

So I assume it is a matter of reflections of the plane-objects. Is this a known bug? How do I get rid of it?

Regards,
Volker

[TheBadger]

looks cool. No idea what the problem is though.

[bobbystahr]

Looks like a tweak of the Index of Refraction to me...haven't run into this yet...care to post the .tgd , I'd like a look at this?....

[Volker Harun]

Feel free to investigate the setup

I tried different settings of the refraction index - from 5 to 10. Made no significant difference.

[Tangled-Universe]

I'm having a look at it now.

There are 2 things I'm wondering about, one unrelated which is the power fractal attached to the surface layer's displacement input port.
As far as I know that input port expects colour/scalar and its multiplier facilitates the displacement, but that's set to 0 so I suppose it isn't doing anything at the moment?
There is displacement in the fractal itself, but then it should be connected to the child layer I'd tend to think.
Anyway, probably not important, because...

At first look I think you're greatly undersampling the reflections. It's softness is set at 0.5(!!) which is huge and is using a mere 4 samples at the moment.
I use soft reflections too sometimes and a softness of >0.02 generally already requires 8 samples. That's with vegetation and such though, so not entirely the same situation.
However, I think this is the issue here.

[Tangled-Universe]

Yep that's the issue...too few reflection samples for this amount of softness.
In the example below I set the softness from 0.5 to 0.05 and it still shows issues, so still not enough samples, but it's far less worse.
I'll see if I have the patience to render a crop which uses sufficient samples for reflection softness @ 0.5 but I think it will be >32 for smooth results and we know how slow TG is with this kind of stuff

[bobbystahr]

whew...never gone up to 32 samples...10 puts my machine in a near coma, LOL

[Tangled-Universe]

It wasn't that bad Bobby, here's how it looks with 32 samples...rendered in about half an hour on a i7-2600k...

[Volker Harun]

Thanks for the testing and advice.

That displacement function had a value, I turned it down to zero for testing.