Greenscreen Compositing?

DeeJco1

Member
I've written a short that involves lots of small puppets in the sames seen with people. I figured this would have to be done by compositing the puppets and people in front of a background.
My questions are:

1. How can ensure the color will match from element to element and all of the elements to the background?

2. How can I make sure the angle and size of each puppet is correct when setting the lens' zoom and framing? (in other words, a puppet in the backgound of the shot should look relatively smaller and blurrier than the foreground subjects.)

3. How can I ensure all the puppets look like they're really interacting with their environment, which won't be in the shot with them at the time, but instead added later during the compositing phase?

I'm using my Canon xl2 and a green screen bought on ebay.

Thanks,

Dave
 
Lots to answer here and I only have a moment so I'll just address the first issue:

Shoot a few seconds of a color chart at the start of each take. In post you can go in and set the black, white, and grey points based on the chart and the colors will match. Matching the foreground colors to the background can be more difficult. Are you planning on using static images? Video of some sort?
 
The backgrounds will be video shot of the locations (an apartment, interior of a car that's in motion, outdoors walking along the sidewalk). All of this while having a small puppet and people in the same shot together.

I know it seems like a job for the team at ILM but with proper knowledge anything is possible.

Thanks in advance,

Dave
 
Can you not film the people in the locations? The less elements to composite the better.
One of the largest challenges with green/blue screen is to match the lighting to the background plate. If you can shoot the people in the actual scene then that eliminates all the associated problems (additional shoot, matching lighting, position, keying)

Since MiniDV has a large depth of field you might consider actually shooting them in the scene at the same time. This is forced perspective. Move the puppet toward or away until it's the right size. If you can shoot from a fixed tripod (which you'll need to do for all of this if you don't want to drive yourself crazy) then you can shoot clena footage (no puppets or people) That way you can matte the bottom part of the puppet if you want to lose an arm or do wire removal easily.

1. Color match- color balance to white for each element. Do final adjustemnt in compositing program by looking to a previous image.

2. To match size create a monster stick or size reference. If the puppet is supposed to be 6 feet have another actor run through the scene and shoot that as a reference for size and placement of puppet. If a lot smaller cut foamcore to match or if taller create a pole the right height.
Use an inclinameter (good hardware store for measuring tilt, just simpel arrow with protractor type thing) use thsi same tilt when shooting puppet. Take measurements of person and standin puppet (monster stick) and height above floor. Scale these measurements to puppet scale. (i.e. if puppet is 1 ft and is supposed to be 6 feet then you'd divid all linear measurmeents by 6)
3. To interact- Be aware of lighting. If light is coming through window at an angle then you'd light the puppet to match and use a card below to try to retain the shadow as well if possible.
Consider rigging somethign inthe shot (i.e. chair) with wire or somethign so it moves. Move the puppet with the same timings and sync up in compositing. These things really make a different.

note that you'll want a monster stick or standin anyway on your run through so your actors know where they're looking. Nothing worse than having an actor looking at the puppets feet or the ceiling when speaking to the puppet.
 
I was looking thru a DV magazine and saw an ad which has a "how it was done" example of a Sin City shot. It shows Micky Rourke in front of the green screen, half way down it shows his alpha channel with the 3d trees digitally added, then at the bottom is the final composite fully rendered out.

I noticed that in the shot of him infront of the green screen, he is tinted with a red-ish light. And I believe that red is the opposite of green so was this done intentionally to get a better key?

If so, once the key is pulled, won't you have to re-introduce more green into the image to get the colors back to normal. Because his shirt is sort of red in the green screen shot but in the composite it is perfectly white. Even though the movie was desaturated to look black and white, his shirt has no hint of gray in it. It's pure white. So was green re-introduced to cancel out the added red from the green screen shot?

Dave
 
Dee, I saw the same ad and thought the same thing. I think once they get it keyed in, they can take the red and turn it into something else based on luma and whatnot - because there were certain things that were pure white in the film - like blood, correct (I'm not certain of this, never saw it I'm afraid, even though I wanted to)
 
Back
Top