"Use Light Cache for glossy": per elements or per object

[cecofuli]

Hi,

"Use Light Cache for glossy" is a usefull to for speed-up the rendering time. If there are a lot af surface with high glossy value (like glossy 0.6 or 07) the difference with pure DMC samplig is minimal. But when we are a low glossy parameter, like 0.95, often i see the "patch" of LC. For example, in my last interior room, for the floor the "Use Light Cache for glossy", and it's great, but for some vases with low glossy, (LC filter is 8 and subdivs 6000 and size 2 cm) i see the samples. And the difference with "Use Light Cache for glossy" is visible (much better when OFF).

So, in my opinion, it would be nice if i can deselect the "Use Light Cache for glossy" per object" or "per material".

Thank's

cecofuli

[Clifton Santiago]

That certainly would be helpful. But there is a workaround: Click Use Interpolation in the material where you want to avoid using LC for glossies. I've found that glossy interpolation overrides LC acceleration.

I often use it for separating glossy reflection vs glossy refraction because I don't like how LC accelerates glossy refractions. It looks totally different than regular or interpolated. Of course, you can't use this trick with animations...

[DanielBrew]

I agree it would be useful. As Clifton said, it's usually the refraction where I tend to get 'patches' where the lightcache samples are.

Another work around is change the filter from nearest to fixed in your lightcache settings. It makes the lightcache much smoother so you don't see the patches.

Dan

[cecofuli]

Hi,

ok, but i want a fulll DMC reflection, not interpolation.

[^Lele^]

Use the material option that's been there from the start: "Treat Glossy Rays As GI rays" set to "Always"
Enable reflective GI caustics in the GI panel
Done.

[cecofuli]

Hi Lele,

thank's for the support, but it' s not the same thing. . And the result is not the same.

bye

cecofuli

[^Lele^]

How is it not the same?
Can you tell which of the two has the option on?
As one is accelerated and the other isn't.
And there's no LC, but the speed is exactly the same as if it was done using the LC.
I'm confused now.

[duke2]

How is it not the same?
Can you tell which of the two has the option on?
As one is accelerated and the other isn't.
And there's no LC, but the speed is exactly the same as if it was done using the LC.
I'm confused now.

Now try higher glossiness, like .95

[cecofuli]

Hi Lele, can you send me this beautifull scene for testing? . Anyway, i don't want to enable reflective GI caustics in the GI panel.

Thank's

cecofuli

[duke2]

To further illustrate (the darkening is caused by the VrayLight set to store with irradiance). There's a curtain behind a sheet of glass, and both are reflected in a mirror to the right. Reflection and refraction on both the glass and mirror are set to .99 to "activate" LC glossies.:







Scene file (Max2009, VRay 1.5 SP2):
http://www.dukecg.net/Glossies.zip

In conclusion, LC glossies look godawful in the 0.9-0.99 range. My only suggestion to the original request would be that it be per-material rather than per-object/element (a checkbox somewhere).

[^Lele^]

Well, turns out that A) they work with the LC, and B) they work per material storing them with the irmap.
LC stored (along with direct illumination) in the first image, and stored per material with the IRMap in the second (lc as secondary engine, but storing neither glossies nor direct light).
Having an LC shading rate of 0.033 (ie 3% LC samples as there are pixels), in adaptive mode, a stored in irmap area light with only 8 subdivs, a high sensitive, low rate IRMap, and GI caustics off means asking a LOT to the rendering engine.
The LC was cured by upping the Samples (sr of 2 or more), no adaptiveness, a rendertime nearest filter of 10 (instead of prefilter of 100).
In both cases, GI glossies were turned on, and the light subdivs upped to 24.
edit: oh, and reverted the DMC noise threshold to 0.01 to boost a bit the overall speed

[duke2]

Well, turns out that A) they work with the LC, and B) they work per material storing them with the irmap.
LC stored (along with direct illumination) in the first image, and stored per material with the IRMap in the second (lc as secondary engine, but storing neither glossies nor direct light).
Having an LC shading rate of 0.033 (ie 3% LC samples as there are pixels), in adaptive mode, a stored in irmap area light with only 8 subdivs, a high sensitive, low rate IRMap, and GI caustics off means asking a LOT to the rendering engine.
The LC was cured by upping the Samples (sr of 2 or more), no adaptiveness, a rendertime nearest filter of 10 (instead of prefilter of 100).
In both cases, GI glossies were turned on, and the light subdivs upped to 24.
edit: oh, and reverted the DMC noise threshold to 0.01 to boost a bit the overall speed

I enjoy asking alot of VRay (although I wouldn't usually use store with IR lights) - it's half the fun! Thanks for the further information, I never really noticed much of a difference between 100 LC samples and 1000, but it definitely comes into play here. Regarding adaptive mode - I thought it resulted in a cleaner solution with less samples with a minimal rendering hit (ie. there's no reason NOT to use it)?

[vlado]

You can get the same effect as the "Use light cache for glossy rays" on a per-material basis if you set the "Treat glossy rays as GI rays" to "Always" for the materials that you want, while keeping the "Use light cache for glossy rays" option itself off.

Best regards,
Vlado

[cecofuli]

Hi Vlado,

thank's for the tips!

bye

cecofuli

[DONGHU_SENIOR]

The shadow of doing this inside-out to those parameters ?