Most of the Ten Before footage was always going to be footage of someone talking in front of a green screen. The idea was to composite this talker(talent) into a 3D environment or room so that various graphics and text elements could be introduced into the scene as if by magic, well sort of.

So we did a few tests, hmmm, from the onset it was obvious that the camera was introducing a lot of edge sharpening. The camera I used was a Panasonic DVX100BE. After some research on the net I found that this was indeed true and that I was supposed to change the detail levels to a so called sweet spot of -3 (see an brilliant forum discussion at http://forums.creativecow.net/thread/193/868436).

I did set the Panasonic’s edge enhancement to -3 in one of the Scene File settings; that is both the Detail Level and the V Detail Level. But I have to say that I still struggled with a halo effect around the talent when I performed the Keying in Shake 4.1. I’ll explain what I did in a different section, but suffice to say – I did a lot of manipulation of the UV channels to get an ok result (bearing in mind the 4:2:0 limitations in this SD PAL camera).

I must say that the Panasonic DVX 100BE is a cool camera for general shooting and that with decent lighting and proper technique on can get a good enough key, but I dream of the day when I can try Panasonic HVX200A or better the HPX170 – both 4:2:2 colour sub sampling cameras, or ummm the errrr coolest of all RED One with 4:4:4 oh yeah.

The two things I insisted on when this Ten Before was shot was that one; the camera must not move and two; the talent must be as still as possible. Purely to make the keying a simpler process. I shot the segments at either a full body shot or a medium close up; filming all the full body shots on after the other then all the mediums. This was a huge time saver because I was able to apply the same key to all the same shot types – with, of course, slight changes to the rotoscope I used as a garbage matte. This worked well, but most people commented that the lack of movement was boring – well you win some and you loose some.

Here is a still from one of the full body shots, you can see that I did not have enough width to work with – another reason that movement was restricted – next time get more material.

The green screen I used was something I picked up at a material shop; very cheap, very bad. But it was a good try none the less.

As far as lighting goes, I used three 150W dedo lights to light Morgan, and then two 2x4ft KinoFlos to light the screen. With all that lighting I achieved consistency during the shoot but I felt that I could have done with more light, maybe more ambient light. The shadows around the legs became a problem in Shake – I had to roto them quite closely; wasting a bit of time and giving the feet an unnatural look. I suppose I could have used the shadows to create a ‘reflection’ but then I would not have had control of that reflection.

Ok I also realized that the lighting of the green screen was not even, I should have used a vector scope. And yes I did use the one in Final Cut Pro 6.0.3 during testing but for some reason I forgot about it when I did the actual shoot. Too caught up in things you know . Next time I’ll make sure the camera is hooked up to the MBP and FCP has it’s scopes open in the capture window (Non controllable device selected).

Right, well, any comments? Next I’ll explain what I did in Shake and FCP along with a few screen shots