A lesson in challenging what you think you know.
We recently finished a project that required us to film four executives on green screen. During our post mortem of the project, Kyle Gustafson (one of our senior editors) said he was surprised by how much work it took to pull a clean key. I was shocked. We had shot on our Red Helium, recorded in 8K at 8:1 compression, and everything was lit nicely. This seemed like it should have been an ideal scenario and simple keying operation. The final composited shots turned out amazing, but it took Kyle considerably more time than we had planned.
He was convinced that we could do better and set out to up our green screen game.
We settled on three key variables that would be fairly simple and effective to change, and test:
- The difference between lighting the green with the same light sources as the subject, in this case, daylight, vs. specialty 525nm Green Lamps from Kinoflo.
- The impact that various exposure levels on the green screen, and specifically, exposing it higher than we normally do.
- The impact of shutter angle on reducing motion blur, even for simple interviews.
A few weeks later and after significant internal debate about what the results would show, we had some Kinoflo 525nm Green Lamps in hand and setup a shoot to test a bunch of variables.
Test #1 – 525nm Green Lamps vs Daylight Lamps
I’ll jump right to the point…lighting the green screen with 525nm Green Lamps made a HUGE difference in how clean of a key we could pull and how much work it took to get there. Why? These lamps are engineered to produce a green color spike at a specific color temperature and frequency, which produces a much more vivid and “green” green than we could ever achieve with a daylight balanced source. In case I’m not being clear, let me say it again, I was shocked at how drastic the difference was between the daylight balanced lamps and the 525nm lamps. According to Kinoflo, the lamps have a narrow concentration of green between 500nm and 560nm, and this is impossible to miss.
Oddly enough, the bulbs are discontinued, but you can still pick them up used pretty easily. Our account manager at AbelCine (our go-to equipment dealer) helped us out and located a bunch of used bulbs, so we stocked up.
Green screen lit with KF55 lamps
Green screen lit with 525nm green lamps
Test #2 – Exposure Levels (of the Green Screen)
The second big surprise was how much cleaner the key was, and how much less work it took to pull a clean key, when the green screen was exposed at a higher level than traditional wisdom dictates. I was taught (rightly or wrongly) to light green screen with skin tones around 60-70 IRE with the green sitting around 40 IRE (on an Rec709 waveform). We shot our tests on a Red Helium and measured exposure using the Gio Scope. With the subject skin tones at stop 13 on the Gio, we tested the green screen at stop 11, 12, and 13, and hands down, 13 gave us the cleanest key with the least amount of time required to pull a clean key. For reference, this is significantly brighter than we would have previously exposed the green. Riddle me shocked, again!
This is also a good time to mention that we also ran tests exposing the daylight balanced lamps at a higher exposure – both 12 and 13 on the Red’s Gio scope. While this did yield better keys, the results are nowhere near as dramatic as with the 525nm green lamps. This does show; however, that my previous school of thought to expose green screen to 40 IRE is darker than current cameras and keying software wants it.
Test #3 – Shutter Angle
Since our tests were focused only on interviews, we only tested 180° and 90° shutter angles. This did provide a noticeable reduction in motion blur during hand gesturing without feeling “off.” Obviously, you need a lot more light output at 90° than you do at 180°, so plan accordingly. When filming interviews on green screen we aren’t dealing with much motion, just the occasional hand gesture, so making changes to shutter angle would require much more testing for more complex VFX shots.
Another element that comes into play on this topic…we often shoot interviews with a teleprompter or an EyeDirect, both of which decrease light sensitivity by about a stop. This means loss of light sensitivity becomes a double whammy. The net result…we need a significant amount more light if we’re shooting interviews with an EyeDirect AND using a 90° shutter angle. Not a show stopper, but definitely a consideration when planning locations and light choices.
The moral of the story….challenge what you think you know and test, test, test!
Any one else have any tips or favorite techniques to share?