I have an internal walkthrough im doing at the mo…render times are getting scary (up to 40 mins a frame at 800x540px) using the irradiance setting for the primary bounce but the outputs are things of beauty. Im trying to reduce the render times as we are trying to get render times down to between 5-10mins a frame. I’ve been playing around with using light cache for the primary bounce and the results arent too bad though the images are quite as good as using the irradiance setting. BUT…render times are halved at worst…only problem is that on the diffused lighting from the ceiling lighting its a little blotchy and reflections are not as strong on some surfaces…see attached images…I was wondering if there was anything I could tweak within the light cache settings to get it up to the irradiance version standard…
The first still is using light cache as the primary bounce…notice the diffuse lighting on the ceiling cast by the celing lights is a little blotchy. Render time is only around 5 mins
The is the same still but using irradiance setting as the primary bounce. Notice how the lighting in the image is a little more accurate when you compare the diffuse lighting on the ceiling..Render time is much more at about 19mins though which isnt good if your looking to do an anim.
Any ideas how I can get round these little annoying problems with the light cache as the render times using this instead of the irradiance method as the primary bounce are much much better…
i dont know really of anyone who has said they use lightcache as primary bounce in a production environment for ‘final’ renders. To me 19 minutes a frame for an interior animation with reflection/refractions and maybe even glossies is pretty good if you like the quality. Renderfarms were made for GI renderers.
Hi, yeh, I know…we dont have a render farm as such (only 2 machines!!) so rendering this out is going to be a slooooooooww process. Some of the frame times can go up to 45mins which means many many weeks of render time will be needed. Would turning of any objects that is behind the camera reduce the render times? I notice that depending on what the camera sees vray reduces/increasing the number of primitives…is this the same idea as turning the objects that you cant see…does vray do this automatically???
It looks like you’re doing a walkthrough with just an animated camera (no object animation) so I’d suggest rendering out an irradiance map for the whole thing and using the saved irradiance map to render out the frames - it should be much, much quicker and produce really nice results. A really quick tutorial to get you going…Chaos: Industry-leading design and visualization software_imap2.htm
Yes, its only the camera that moves though some doors open as your walking through the areas…would this create problems? I’ll take a look at the tutorial…this could be what Im looking for…
you really need to post all of your settings if you want others to help get your render times down…that is unless you just mention everything you’ve changed from default values. the single most important variable to speed vs quality (IMO) is image sampling, which you didn’t mention. did you play with the noise threshold?
it is possible to animate doors in your scene with a saved GI map, you just have to do it carefully and be able to live with certain things. you can try disabling the generate gi option on the door, you can mask the view of shadows and the bounced gi with other objects, you can bombard the doors with extra samples during the time in which they’re animated,…instead of doing composite…it’s not ideal but there are workarounds that can work in some situations
