|
|
|
|
|
by atoav
1418 days ago
|
|
As a VFX freelancer who is now in education I really want to stress the value of optimizing your renders. If things are far away, prerender them and replace them with billboards, if you have a lot of duplicate gemoetry clone them using alt+d instead of shift+d or use geometry nodes/particles. Think about how many samples must be used where. Do test renders with the workbench renderer to judge the speed and timing before doing the final thing. The energy we put into animations is real. Going from 150 samples to 200 will maybe reduce the noise in a tiny portion of the image a little bit, but the energy and time used for it is real. After a certain threshold every kWh you spend is spent on very subtle improvements. With my students I can typically reduce render times by factor 10 just by selecting the right settings and using details where it matters instead of everywhere. A render that would take two weeks on their machine would then be done in 36 hours. Independent of the resources you use (e.g. a crowdsourced renderfarm) these optimizations should be done. |
|