LC+IM -> LC pass much darker than IM - light portal bug?

[Micha]

Hallo,

I'm a Vray for Rhino user and have some times a problem, that I have seen at Vray for Max in the past too.

My problem is, that the LC pass looks much darker tha IM and final pass. My setup is:

Secondary engine - light cache 1200-0.01 - GI multiplier 0.8
primary engine - IM
interior with two arealights in portal mode and "store with irradiance" map

I tested GI multiplier 1 and to render without "store ..", but nothing helps. Any ideas?

Ciao Micha

[percydaman]

I'm not following. Your LC and Irrad map passes wont necessarily look the same, while they are calculating, as your final pass. Its nothing to worry about.

[ene.xis]

maybe ur unticking the store direct light option in the LC settis?

[travistlo]

The light cache in your case is set to secondary bounces. These are the bounces after the first initial light bounce. That why the Irr. Map is brighter, the Irr. Map is set to calculate the first pass which is brighter than the secondary light bounces.

[Micha]

So, I can not use the LC pass as a quick noisy preview of the scene and adjust my lighting. The LC show the direct light (it's enabled) and bounces after the first initial bounce. The first GI bounce is not visible. Interesting. Thanks for the answers.

[percydaman]

try using LC for both first and second bounce.

[themaxxer]

try using LC for both first and second bounce.

I second that.

[Micha]

Both at LC? And should I enable PPT?

[Micha]

Next round. The answer seems to be not so easy: both engines at LC in PPT mode. I cut some window in the PPT rendering and we can see the previous IM/LC/Final passes. New riddle: which passes are wrong exposured?



EDIT: ... a general question: are emitters or arealights better for PPT?

[mmikee]

Dont check ppt for test renders. Just use LC/LC.
Personally, I just use the ir/lc and turn down the other settings for testing.

Reards,
Mike

[travistlo]

New riddle: which passes are wrong exposured?
PPT?

None of them. They are just all showing differnt things.

The first bounce, and the secondary bounces have to be combined to get the final lighting. Thats why they look differnt. The PPT still has alot of black pixels that havent been calcuated, which can make the image look darker than the final render.

For test renderings, Use the technique that you will render your final shot (Irr map & LC, QMC & LC, whatever) Just turn down the settings. The Shadows may be grainy, or they will be splotchy whatever, but the overall light value should be more accurate than some of the other methods.

[Micha]

I think, it's a bug. So, how could we evaluate, that it is a bug or not? It should not be, that an rendering based on PPT/LC looks totaly different than based on IM.

Here my last test: 1h33min PPT and IM based on the lich cache file of the PPT rendering. It is allways the same - the rendering based on IM is much brighter and contrastful. Vray for Rhino base on vray 1.49.73. I use the universal method with Adaptiv QMC 1/100 and noisethreshold 0.01. The window is an area light portal (no store with irradiance map).



EDIT: If I disable the light portals, than I get with IM the same brightness like with PPT. So, the light portals cause the wrong exposure in IM mode.

[Micha]

Here my final shot without the light portals.


Can it be avoid that light portals destroy the natural lighting? I like light portals for smooth and fast IM based GI renderings.

[percydaman]

well you may be right, i hardly use light portals because I dont care for them.

[studioDIM]

I think, it's a bug. So, how could we evaluate, that it is a bug or not? It should not be, that an rendering based on PPT/LC looks totaly different than based on IM.

I cannot speak for the rhino version, but it IS normal that different methods lead to somewhat different results.
LC make NO distinction between primary or secondary. When used LC calculates both at the same time.
The display of the LC calculation pass is a generic reference for the user to spot light levels, not a finished, interpolated calculation.
The usage of a different primary method is there to give the user more control over the detail of the solution.
Because of the way GI sampling works (whether it is Irradiance mapping, or qmc), the higher the QMC samples (call them subdivs if you please), the brighter the image will be.
In normal conditions, this is hardly noticeable; however, under particular circumstances, raising the sampling rate will add in a visible amount of light (and accuracy, of course).

As for ppt, i think you are still a bit confused.
PPT is progressive path tracing, and it leads to a finished image. there's no saving a ppt cache file, much less so reusing it for LC/irmap renders later on.
Save an LC cache file, and use it with LC later.
Or render an image using ppt, and let it run until it's complete-satisfactory.
PPT overrides any other method, AA setting and so on.
Think of it as the Max**ll method of rendering.
It's likely that in your sample above, you lit the scene TWICE.
Lighting levels seem precisely doubled in the second image.
So you loaded a prelit scene, and then added on IM.

To recap, make sure you get the rendering settings right, and refer often to the samples presented in the vray online manual.

Hope this helps,
Lele