Well I’m working on an animation on Blender, but the Shadow samples are making it go 10+ minutes for each frame!
But if i lower them i get a very very grainy render. I’ve set the Threads to 8, but still very very slow, is there anyway (I’m shure there are lots) i can go around this? Because I’m really pressed for time (I’ts for a client) and these render times are really a no-go. :eek:
Thanks
Edit:
The scene is and interior workshop, with several windows, and no indoor light except for one room. I have a sun with 5 samples, a area light with 4 samples with orange color, and several blue area lights with 5 samples on the windows.
Reduce your samples back to the default of 1.0. This will speed things up. Threads are the number of processors in your system. So increasing them is not going to really make it go faster unless you have an 8 core system. In that case, why would you lower it?
Separate out your shadow pass and do a slight blur on it then use a mix node to add it back into the final image. this way you can recover some of that soft shadowing you are trying to achieve with 4 samples on your lights.
Also AO in approximate mode can give you a substantial speed increase. You can just up the error value to make it render faster.
These suggestions will alter the look slightly. If the client has already seen/approved your other look, you may be stuck and just have to wait it out. But if you make it look better using some of these suggestions perhaps you can get sign off?
If you’re ready to branch into the world of node-based rendering, you can do a lot about those shadows.
First, set up your scene with no shadows at all … turn them all off. Go for great, even illumination of everything, highlights where you want them, and … no shadows.
To get those characteristics, just the way you want them, group things into render layers and deal with each thing separately … no shadows.
Now, add nodes to the mix which deal exclusively with shadows, using “shadow-only lights” aimed at materials which are also “shadow only.” You are now dealing with a channel of information that, if viewed in isolation, consists of variously-dense and maybe variously-colored shadows on a “blue screen.”
The last step … really a 2-D step … is to “mix-down” these channels of information to get just the effect that you want. The shadows need to be consistent, both in position and in density, with “what the lights would have produced,” at least enough to be plausible. But they’re a separate channel and so you can adjust them with “the usual knobs.”
I started doing this in my earliest days of “moving beyond the ‘Big Fat Render Button’,” when I noticed that a very disproportionate amount of (re-)render time was being (wasted) … not because of getting the light right, but because of getting the shadows right.