Question on lighting for chromakeying

[Bob Woodhead]

First off, do some Googling, and you should be able to find some images of lighting plots for chromakeying. A picture’s worth…. you know. Anyway, since you’ve not shot greenscreen before, I’d suggest if at all possible to NOT attempt shooting your talent head to foot. The transition zone at the feet, and shadows along with them, is by far the most challenging aspect of pulling a good looking greenscreen. If you can avoid using a full length shot, your life will be much easier.
Here’s my checklist for greenscreens (or blue):
– separate talent from screen by 3 to 6 feet (removes problems of chroma spill from screen onto talent)
– pull camera back as far as possible from talent & zoom in to frame shot (especially on 2/3″ chip cameras, but even on small cameras, this creates the out-of-focus background effect, which on a greenscreen, helps smooth out any imperfections in either screen or lighting)
– light talent first. don’t light “for greenscreen”, just light the talent the way that’ll best match the background you’re going to key in
– light the screen. EVEN LIGHTING. Make the lighting as even and smooth as you can. (Did I say EVEN LIGHTING yet?) I love using Lowel Tota lights for this. Usually 2 500W or 1K Totas, 1 on the ground pointing up and 1 on a boom arm or somesuch aiming down. Sometimes I fill in any gaps with a couple of 300W fresnels. Put diffusion on everything. This could be a good place to use “alternative” lights; work lights from Home Depot, etc. Just use the SAME color temperature lights on the screen. For the screen itself, there are some on eBay that are 10×16′, green on canvas backing for about $80 or so. Not true “chromakey green”, but with good lighting it keys great. DON’T “blow out” (overexpose) the screen – you don’t want to wash out the color information.
– you can create a great looking light plot with a single flourescent box, a reflector, a 300W fresnel & a 150w kicker fresnel. Total maybe $650 (w/o stands, arms, etc). I’ve used KinoFlo Divas, I’ve used a box I made myself, and there’s not really much difference (other than the obvious durability, flexibility & looks). There’s a vendor making 4 lamp Diva-knockoffs for about $300… I’d look at those if you’re on a budget. Get one of those big ovoid reflectors that have gold, silver, white, black and translucent sides for about $50. Get a 150w Lowel ProLight as a kicker on the talent and a 300w fresnel (Arri, LTM, Mole) for bouncing off the reflector onto the talent for fill. You might be able to get away with just the reflector bouncing the softlight back as fill, but not always. Total budget should come in around $1K.
– if you have to shoot talent as a WS, more lights, more controls, and better software for keying will be required

Good luck & have fun! Do some tests before the “real deal”.

“Constituo, ergo sum”

Bob Woodhead / Atlanta
http://www.CoolNewMedia.net
Quantel-Avid-FCP-3D-Crayola

[Bob Woodhead]

Just wanted to clarify that while there’s places to save money (work lights to light greenscreen), you’ll come out ahead if you buy used fresnels to complement an inexpensive flourescent softbox. The easiest, best-looking lighting comes from soft sources. That’s one reason flourescents have taken off (and they use much less power). Fresnel lights are worth the investment over open-face units for their softer light (over open-face) & their flexibility. I’ve a Lowel Omni kit that goes with me along with a bunch of fresnels. Guess how often I open the Lowel kit? Not nearly as often as I’ll reach for a fresnel.
Other thing I wanted to repeat was to always think about ways to bounce light, thus “multiplying” your investment.

[Alex Huber]

Hey Bob…

Do you know who is making the Diva clones or where to locate them?…

Thanks,

Alex

[Bob Woodhead]

coollights.biz
Now the guy is working on “democratizing” HMI technology for us as well!

BTW, I don’t own any of his stuff, so I’m not vouching for it – only that I’ve made an even cheaper version of a “Diva-style” unit, and find it gives satisfactory results.