Describe RyanSmith/Homework1 here.
I did the scene setup process manually. I did some math by hand to setup cameras and spotlights that have the "look at" pattern. Typically i would pick a radius and do sin/cos math to find out where the camera or spotlight should be.
For area lights, there was a lot more hand tuning because they don't use the look at pattern. I did a lot of incremental modifications to make sure my sense of coordinate frames made sense. It's convenient that you can see the area lights.
I reduced the number of samples per pixel (sometimes just 1, mostly 2) and per area light to make the render/check cycle take less time. All my test runs were done with just 1 subdiv on the face model. Reducing the resolution of the image did make things faster, but it was too difficult to see if you were getting a good result.
I also had a pretty good workflow running Photoshop on my PC and looking at the images though PC/AFS, instead of running exrtotiff all the time.