For the filming, as a makeshift green-screen I used an orange sheet, stretched out to be creaseless so as not to cast any shadows, lit by various lamps aimed at the actress from all directions. Proceeding through the motions the filming was over relatively quickly, leaving more time for the exhausting process of roto-scoping. But first, came the keying. This is where the first problems began, as ever!
Using After Effects for keying out the background was relatively simple. There were discrepancies here and there, but feathering the edges down and individual selections got rid of anything that looked somewhat out of place or wrong. Using VirtuaDub I exported the completed moive sequence out as individual frames, which turned out to be…a whoooooooooole lot of frames. With a runtime of around 1 minute, working at 25fps, well 25x60x60 = one heart attack. Luckily I restricted movement as much as possible during the filming, so corners would be able to be cut in regards to how many frames I would actually have to draw over, but the number was still pretty daunting.
Taking the frames into photoshop, I began roto-scoping various test shots, trying to see which type of colour and style would actually work in motion. In previous works, I tend to get very addicted to the finer details and line –art of it all, which was something I was being pushed to give up for this process. I quickly learned in order to get a fluid motion between the frames, the detail had to be either incredibly precise, or deliberately basic. I tried the former at first in a couple of test shots, but the amount of work that went into a single frame was just not viable to be doing for each individual frame. With the latter I became utterly unenthused, as this project I had imagined just began too look so scrappy and essentially not truly showing the amount of work and forethought that had been put into it.
I should explain, the example below was the worst test frame I worked on, yet still took me a few hours, only to then botch it up by merging layers and messing up a lot of opacity and masking options, the blue background was a temprary solution which ended up making the picture unusable. To me, this problem alone spoke volumes about the future of the rotoscoping, with every mistake or wrong line making a huge impact on the entire project, leaving little room for error and not enough time to realisticly do it without error.
I'm glad I took this direction for the project as I really wanted to embark on something completely new which would challenge me in new ways, but for now I have to concede victory to rotoscoping. Instead i'll be going back to one of my earlier ideas of making a comic. It will have
the same expectations of expressing and displaying the character, allowing the reader to easily and quickly assess what sort of character she is, purely by her actions, reactions and expression. I intend to create the comic in Flash CS4, aiming to give it sounds, possible dialogue, ambient music and overall with a hint of interactivity. I think this is definitely the direction to take, being more realistically achieveable and not so much of a niche.