






After much planning and experimentation, we realized that we don't really need to use a green-screen for the head removal. As the motion is only along one axis, and the camera is fixed for this shot, we can remove the head in post-production using a simple mask.
For this test, we used ping-pong balls as motion trackers for the head and the broom. We shot the head without any green screen or mask. I had my head turned up so that my head wasn't casting any shadows on the neck and chest. We also shot a footage of just the background wall so that we can use it to mask out the head.
For the broom we hand-held a green poster paper behind it and walked it across the wall, so that we would get the exact lighting conditions.
Later using After Effects we tracked the ball in both the shots and composited the broom over the head. We used a simple garbage mask to clear up the noise on the green screen for the broom. After Effects does not support attaching masks to trackers, so we found this script from AExtensions that does that and used it to move the garbage masks.
For the final shoot we plan to use 3-4 trackers so that we can get head motion, rotation and swivel data. We plan to build a tree branch prop that would be used in place of the head.
This was perhaps the biggest challenge yet on this project. Since none of us really have any experience building prosthetics, it was an incredible learning experience. It involved building the cast with plaster, connecting a 5' PVC tube to it and then mixing/cooking the gelatin, pouring it to create the cast.
This is what the gelatin appliance looks like.
And here is an interesting arrangement we used to pour the gelatin and erect the cast. With nothing more than random found objects in my apartment. Epic indeed t'was!
What you see there is a Styrofoam head with a tube drilled into. Dunked into a plaster cast in the blue bucket. Fitted into another PVC tube. Standing upright using a crutch!
We experimented with casting appliance in Gelatin using plaster and it turned out to be a very successful. The gelatin works very well on skin and has organic sub-surface scattering properties. Will post pictures of the gelatin soon.
So the next is to build the actual cast for the head and make the gelatin parts. Here you can see the Styrofoam head with the clay added to create the positive for the mold.
The idea is to use a 5' PVC pipe to cast a solid gelatin tube with the bald cap at the base of it. Then apply this appliance over a latex bald cap and then blend the edges with more liquid latex.
After much testing, we found that the Palmer body paint works best under our lighting setup. An even mix of about 1:1 paint and water applied with an air-brush, creates a nice, clear, uniform wash on the skin. It hides the skin enough to cover up all the specular highlights, but it is transparent enough to let the muscle, veins and skin texture show up through it.
In the images you cans see the red-tempura paint applied on the left side and the white body-paint on the right side. Clearly the white color responds better to the light and creates a soft diffuse surface.
We've ordered the same paint product from another brand -- Kryolan's which is a bit cheaper. The paints should be here early next week.
HITOTZUKI KAMI & SASU 'OMNI' from Louis Mitchell on Vimeo.
A video that was linked from the Vizagogo 18 email. The colors they use are reminiscent of the palette we plan to use for our video. Its also interesting in how they are using this certain palette. How they are treating it like a friend that is accompanying them on a long journey, a beautiful way to look at color.
One thing we realized in the tests was that even if we got naked models with body paint, the bald caps would never look realistic. After much debate, I decided carefully to alter the design of the character so that it would be more practical. Here's what i came up with. It's a sort of a lotus bud like formation from the top. We could do this with a bald cap, a swim capo and a funnel-tube and some plasticine or some kind of polymer based molding/modeling material.