OTOY Forums

Simple question I hope... Does Octane take into consideration/calculation lights (emition type) even when they are not seen in the scene?

I have a project where I have multiple rooms all in one scene. They are completely isolated from each other with sealed walls etc. However if I change the sampling of lights in one room they clean up much faster....then I move to the other room to render it and those now are falling way behind in cleaning up....so I up their sampling rate and so on....

I love that the sampling rate is in the OBJECT surfaces but for this reason if the above is true....it's hard to prioritize in a big scene like this what I want to have priority, because the sampling priority get's saved in the object. In Native LW I would just save out final scenes to render called, Room1 and Room 2 with adjustments to each for things like this.

Can anyone (mainly Dev) confirm or deny this operation, and could it be addressed?

ok I can confirm with a test that lights in another room that are not visible will AFFECT the sampling rate of other lights. Notice in these two images that I actually have more samples in the more noisy image (which has the lights in the other room (just past the welcome sign) all turned on. But you can see in the cleaner image I don't have as many samples and a similar sampling rate and it's much cleaner. This means that with the other lights ON they are getting focus but are not being seen....

I hope this can be addressed with a quick update. I do not want to have to create different scenes and models for these kinds of projects just to keep the render times down. Looking forward to any feedback.

Thanks for publishing this test. Lighting in Octane is definitely not perfect.

I know breakout models and breakout scenes are a big nuisance. I have to do it quite frequently for the most bizarre reasons. Looks like in this case you have no choice but if someone else can explain a better solution I'm looking forward to seeing it.

Normally in a case like this I would move the camera around over time (I realize you may not normally be doing animation) and then apply an envelope curve in Graph Editor to the Emission nodes' power or sampling setting so only appropriate lights are on when a camera is in a certain room. However, that still means your objects will be scene specific...

hdace wrote:Thanks for publishing this test. Lighting in Octane is definitely not perfect.

I know breakout models and breakout scenes are a big nuisance. I have to do it quite frequently for the most bizarre reasons. Looks like in this case you have no choice but if someone else can explain a better solution I'm looking forward to seeing it.

Normally in a case like this I would move the camera around over time (I realize you may not normally be doing animation) and then apply an envelope curve in Graph Editor to the Emission nodes' power or sampling setting so only appropriate lights are on when a camera is in a certain room. However, that still means your objects will be scene specific...

That's a very interesting idea! We mainly do animation but it's usually scene based, meaning I will move around in this space and then move to the next space...sometimes including a transition piece where we go through the doors etc. But not always. It's just easier to have everything in one file/scene when working with textures etc...keeping things neat and similar.

I hope Juan can comment on any possibility of this being addressed in any other way. Thanks for your comments and confirming your own experiences.

You know that the lights are not visible because you are a human that knows the project, and you know that the lights in the other rooms don't light the current room ... but a path tracer needs to sample all the light emitters available in the scene to know if they are seen from this room or not. For example a small window in a wall could make possible cast in your current room the light of other rooms with a little change in the scene, hard to detect by the render engine.

You are going to need to add priority to the lights in your current room using the sampling rate parameter, or disabling the lights of the other rooms. Perhaps you could build a function to set this parameter using the distance from the camera to the light, or other tricks to automatize the scene.

Anyway lets see if the Octane developers can improve the lighting features in Octane 3.

-Juanjo

juanjgon wrote: You are going to need to add priority to the lights in your current room using the sampling rate parameter, or disabling the lights of the other rooms. Perhaps you could build a function to set this parameter using the distance from the camera to the light, or other tricks to automatize the scene.
-Juanjo

That's a great idea, except rather than using the camera distance to trigger the lights, I would use a null with a distance trigger applied to the lights. The null could be moved to the center of each room as a kind of light switch. Which ever room the null is in, that's the one with the lights on. This would make the models scene independent.

We'll yes I am human...we can agree on that! lol. However I didn't think a path tracer worked as such... if a ray hits a solid wall it will not affect other elements...and I guess that's what your saying...is the other rooms lights have to be sampled and once they hit the wall and don't affect this room they are done...but we had to calculate that to find out! I get it now...that I talked it out. lol.

Building a node like you mention is over my head but I would love someone that could put some effort into that. That might be a great rig to have in the mean time. Thanks for the feedback. Will break my scenes up when needed.

interesting topic! my question at this point is: in octane, is a light set to power=0 totally ignored by further processing or does it still have influence on the general light calculation?

markus

3dworks wrote:interesting topic! my question at this point is: in octane, is a light set to power=0 totally ignored by further processing or does it still have influence on the general light calculation?

markus

It seems in my quick test that setting power to 0 does get rid of most of the calculation but maybe not every bit of it. I could not get the same samples per sec with them set at 0 as I could with the node unplugged. But it was very similar.

Dev or other might reply...