So I was waiting on a render of my Rolex Yacht-master II the other day, and having read a few posts recently by Dave Jerrard on Facebook about his monstrous 32 core workstation; I commented how a little more horsepower would be nice. The subsequent conversation with him was very interesting in terms of the practical outcome. First, a grab of my render in progress (check out the time remaining!)
[image_frame style=”framed” align=”center” height=”200″ width=”600″]http://www.scorpiocgi.co.uk/wp-content/uploads/2012/08/rendering.jpg[/image_frame]
Now I know these estimates are not accurate, but none the less when you are waiting on a render, it’s depressing to see. The settings for that render were as below.
[image_frame style=”framed” align=”center” height=”200″ width=”600″]http://www.scorpiocgi.co.uk/wp-content/uploads/2012/08/settings.jpg[/image_frame]
As you can see Final Gather radiosity was set and interpolated with settings that had worked pretty well on my vehicle renders. It was suggested that what I might want to do would be to use Monte Carlo instead, but WITHOUT interpolation, and with the RPE set really low around 4 or 5. Additionally, given the model I was rendering, the shading and lighting samples while not OTT at 4 would be fine at 1. i also took the opportunity to remove the secondary bounce and to stay with a single bounce.
These were the revised settings.
[image_frame style=”framed” align=”center” height=”200″ width=”600″]http://www.scorpiocgi.co.uk/wp-content/uploads/2012/08/05_lightwave_rendering.jpg[/image_frame]
The maximum samples value on the camera tab is one that needs to be tested and tweaked for your specific render, but I chose just to stick on 128 and get my render going. What was the upshot? Well the render came home in 3h 22m 53s rendering at 3000 x 2000. Pretty good all in all I thought. A few more minutes on top to get an AO render and a depth pass. Job Done!
[image_frame style=”framed” align=”center” height=”200″ width=”600″]http://www.scorpiocgi.co.uk/wp-content/uploads/2012/08/3dartist_final_grade.jpg[/image_frame]
So when I thought interpolated must be the quickest way to go is not always true, and in this instance was very much slower.