Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Performance of deferred lighting?

Discussion in 'Editor & General Support' started by frosted, Aug 31, 2014.

  1. frosted

    frosted

    Joined:
    Jan 17, 2014
    Posts:
    4,044
    Can someone point me to an outline on the performance characteristics of unity's deferred lighting?

    I've read that deferred rendering is roughly o(light sources*pixels) - but clearly this doesn't apply to Unity's deferred lighting model (which seems sort of logarithmic). The only thing I've found is this http://www.realtimerendering.com/blog/deferred-lighting-approaches/ which is a biiiit past my pathetic math skills.

    What I'm really trying to assess is how different scene compositions, renderers, meshes and lighting affect deferred lighting performance, as well as what points deferred lighting exceeds forward rendering with fullforwardshadows.

    I'm mostly dealing with fairly simple scenes for desktop (like 200k-300k triangles on screen), but I'd like to use some more complex lighting, ideally things like multiple real time shadow casting lights. So far I've found that forward lighting is simply faster even with multiple real time lights producing shadows (large radius point lights/spotlights projected onto fullforwardshadows receiver). I'm trying to understand why forward rendering has been faster, and at what point it really makes sense to switch to deferred.
     
    Last edited: Aug 31, 2014
  2. frosted

    frosted

    Joined:
    Jan 17, 2014
    Posts:
    4,044
    Anyone?