Is it slower to render passes separately?

[Anssi Kuha]

Suppose you have ten lights in the scene. Is it significantly slower to render the contributions of each light separately? I dont think it necessarily would have to be, since the combined render will still need to calculate the same “trajectories”? But in practice it probably is, right? How much slower? Not ten times slower, right? Thanks

[Dwayne Savage]

That’s a tricky question. There are some factors that change the answer. Like which render engine? What are your render settings? Like samples, and  bounces. Is there volumetric? Hardware specs? Will it be 1 light per pass or a few lights? 

Really the only true way to know is by testing it with your specific setup. In general it will take longer to sperate the lights into separate passes, because of how Ray tracing and path tracing work. Most of the render time is on the calculation of rendered objects and their materials. Of course emission materials in cycles counts as lights. If you have a large(relative to your hardware) number of light like 50 or more and low end hardware then it might help to break up groups of lights into passes. 

[Omar Domenech]

I think I have heard that enabling that doesn’t really change render time. But that is easily testable. Make a scene, render it one way, see how long it takes, then render it the other way with the lights separately and see if it had an impact on render time. Good old scientific method. 

[Author]

Posts