GISD (Global Illumination Surface Detail) settings test on objects

[Dune]

Interesting test, Kadri. Matt and Jo are probably working hard to improve GISD. Basically it's a treat, but like said here, you have to use it prudently. You can see the other issue addressed as well, the dotted black line on the edges of objects. That should go in the end, I hope they're able to do so.

[Kadri]

Indeed, lately I've been fixated on Stained Glass but fear it's not doable how I want till we get full grayscale transparency....

It is usable but we need better transparency for sure.
I haven't tried that but shouldn't that be possible in some way anyway now?

You know more then us Ulco
So far i know GI kinda solutions does have those kind of problems in other software too.
Mostly long render times are the solution as it looks.
What Matt and Jo is trying hard is probably how to do this in a reasonable render time without much errors.

Osyhan i thought about that but render times made me afraid to try and wanted to test only the options you would say.
I am rendering now with higher setting.
I begun with very high numbers ( GI cache detail 8, GI sample quality 8 ) just to be sure and will go lower from there.
Just the GI pass takes on this 1280x1000 more then 20 minutes with this setting.

[j meyer]

Don't know how many polygons your object has,but I would test it
with a higher polycount,too.Just to compare the blotchiness of the
results.If there is no difference you can at least exclude the polycount
being a cause.

[Kadri]

Yep that is one of the things i want to try too.
Every wall is only one big polygon.

Looks like by going high to "GI cache detail" 8 and "GI sample quality" 8 i found the setting that in this situation got rid of the blotches.
By using values 6, 6 the blotches could still be seen.

I am trying with a value of 6 on two of them with GI blur radius 16 ones more to see what happens.
I stopped the render because the blotches are still there.

By the way, i saved as an EXR too to see if there was any difference but it had the same blotches (with 6,6).

This test is kinda an extreme situation.
With a texture on the wall and displacements it should be barely noticeable i think.
Maybe even with lesser numbers.
And with an animation the interpolation would help too probably.

[Kadri]

J meyer i tried it with a very high poly object too but there was no difference.

This is so far i will test it if there aren't other suggestions guys.
Hope it helps a little more to understand some of those settings .

[bobbystahr]

J meyer i tried it with a very high poly object too but there was no difference.

This is so far i will test it if there aren't other suggestions guys.
Hope it helps a little more to understand some of those settings .

Try it with the Rock tgo with it a fair bit of PF displacements to see the strength of the effect....

[Kadri]

I tried it with this kind object because on such an object it might be much harder to see.

[bobbystahr]

I tried it with this kind object because on such an object it might be much harder to see.

The reason I suggested this approach is I viewed a video tute on Ambient Occlusion by I think Oshyan that got me insipired to play around with the setting.
Dunno where it was as I dl'd it and found it on my HDD....