PDA

View Full Version : Question regarding portal lights LW 2018



madno
07-09-2018, 04:39 AM
Hi,

I always thought portal lights only gather samples from the environment in case GI is used.
If Normalize is off and Intensity set to 1 then the environments intensity is transfered 1 to 1.

I now noticed that portal lights emit their own light, even if GI is off.

Is that intended?

GI off and no other light present.
142157

Tobian
07-09-2018, 05:52 AM
Portal lights *are* lights, they serve to replace sampling of the environment in a radiosity solution, as you get a cleaner result from smaller importance sampled lights, than you do with a GI environment sample or a Environment light. The trick with these lights is to remember to disable 'sample background' in the radiosity tab, or it will introduce both direct and indirect sampling from the environment. If you want to isolate the light from these lights, you can always make them a light group. You should get much cleaner radiosity solutions from the portal lights, than you do with the sample background setting in radiosity both because it's much finer importance sampling, based on area lights, and also because you could do interpolated for the main bounce, but lights work more like brute force, which means you should get way less blotching issues.

Andy Webb
07-09-2018, 05:57 AM
Portal lights *are* lights, they serve to replace sampling of the environment in a radiosity solution, as you get a cleaner result from smaller importance sampled lights, than you do with a GI environment sample or a Environment light. The trick with these lights is to remember to disable 'sample background' in the radiosity tab, or it will introduce both direct and indirect sampling from the environment. If you want to isolate the light from these lights, you can always make them a light group. You should get much cleaner radiosity solutions from the portal lights, than you do with the sample background setting in radiosity both because it's much finer importance sampling, based on area lights, and also because you could do interpolated for the main bounce, but lights work more like brute force, which means you should get way less blotching issues.

Well that is a great piece of information, something to remember. Thanks

madno
07-09-2018, 06:17 AM
Portal lights *are* lights, they serve to replace sampling of the environment in a radiosity solution ...
Many thanks for chiming in. Most of it I understood already from the manual and online sources over the time.
But, if portals are for GI sampling and GI is completely disabled, why do they still create light? That's the point where I am confused.

gerry_g
07-09-2018, 06:42 AM
GI samples the backdrop, it’s light input is from this and not scene lights ( though they can add to it), a portal light samples the backdrop so it is one and the same, a full replacement for GI

Tobian
07-09-2018, 06:53 AM
When you use a light as a portal light it takes a spherical environmental sample, just like the env light does, or the radiosity BG sample, so you're just substituting one sampling of the environment with another. The difference is only in which way it is done. With the radiosity sample, it is taken in the indirect diffuse sample, so it appears in the 'diffuse indirect' buffer. If it's done as a 'light' either in the forum of an environment light, or a portal light, then it's still a light, so it appears in the diffuse direct buffer.

I did a video explaining the different areas you need to look at for the Env light https://www.youtube.com/watch?v=72AAwSFx4nA the information is exactly the same, only for interiors portal lights give much cleaner/less noisy results than an environment light. in both cases using these lights is a substitute for the sample BG tickbox in radiosity, and in many cases it will be slower, however it should be a lot cleaner, and more accurate. If your environment has an HDR, which has a bright sun, you will get sunlight, and hard shadows as well as the background fill from both an env light and portal lights.

madno
07-09-2018, 07:14 AM
Ah, now I got it. Thanks for explaining it again.