OT: Blue Screen vs. Green Screen

Hammer65 wrote on 4/13/2005, 12:57 PM
Was reading this article about Battlestar Galactica and Star Trek : Enterprise using a HD capture instead of film to save cost. In it they talke about doing green screen work and having that process worked out fairly well but not blue screen. I never realized there was a difference? Why would you chose to use a green vs. blue for composite shots? I can't see why it would make a difference?

Here's the article, not a bad read.
http://millimeter.com/mag/video_breaking_comfort_zone/

Comments

Spot|DSE wrote on 4/13/2005, 1:05 PM
If you're using high compression Digital, you almost always want green for it's luminant value. There are several articles on this on the web, including 3-4 on th VASST site.
For lesser or uncompressed vid or film, blue works well too.
for an experiment, create a color bar on the Vegas timeline. then drop the Black/white filter on it. Look at what color is most dominant outside of white.
That's why we use green.
riredale wrote on 4/13/2005, 2:47 PM
The way I interpret it is that, in the DV format, luminance is represented by 480x720 pixels (for NTSC), while the color information is represented by only 480x180 pixels. It's slightly different for PAL, but you get the idea.

So in order to get the cleanest edge definition, you want the highest spatial resolution. Luminance is mostly green, so green gets you the best results. Of course, you have to make sure no one is wearing green in the shot.
rmack350 wrote on 4/13/2005, 4:59 PM
The problem with the logic is that green is still a calculated value for every single pixel. Not that I want to argue about it, but to get a green value you still have to use the red and blue values to derive it so you still should just get a concrete value for green every 4th pixel. All others are interpolated.

This is for NTSC DV, of course, where red and blue are onle sampled every 4th pixel. For Star Trek, etc, they have much better color sampling.

Off to read the article...

Rob Mack
Hammer65 wrote on 4/14/2005, 11:31 AM
Thanks for the feedback. I guess it begs the question then why not just use green all the time unless you have green in the shot. They never say why but imply a need to use blue screen. Being a novice I thought it was just a matter of looking at each pixel and replacing pixels that match your chosen color, be it green, blue, red, etc. Sounds like green works best for DV, does blue work better with film?
rmack350 wrote on 4/14/2005, 1:37 PM
Well, it's a really good question that would bear a bit of research-whether blue is actually better for film or not.

The quotes in the article say that blue was bad for digital aquisition because of compression issues-which is not neccessarily the same as color sampling issues.

Not trying to argue here but I'm wondering how much of the reasoning behind green Vs blue is lore instead of fact. And I recognize that green could be better but perhaps the reasoning isn't right.

Rob Mack
riredale wrote on 4/14/2005, 5:40 PM
I don't have a reference website handy to pull out here, but just imagine using a blue screen with DV. In NTSC countries, luminance pixels are spaced out over 480 lines with 720 pixels per line. The blue information is on 480 lines (the maximum possible vertical resolution) but is colocated with only every 4th luminance pixel on any line (really horrible horizontal resolution). So a bluescreen key using DV would show nice sharp transitions in the vertical direction (horizontal edges) but very poor keying in the horizontal direction (vertical edges).

PAL countries use a different color-sampling pattern, so the bluescreen horizontal resolution would be better at the cost of the vertical resolution.

I'll have to dig around for that website that shows all this stuff...
rmack350 wrote on 4/15/2005, 7:44 AM
I'm not arguing that but it just seems to me that you can't actually see a green value unless you use the red and blue values to derive it from the luma. You can say that luma includes green but luma is only brightness. You don't actually get a green value unless you calculate it using the red, blue, and luma.

So it seems to me that green values in between every color sample will be just as approximate as blue or red values for those pixels.

Look, I've lit greenscreen, we've pulled green screen keys here from what I've lit. They looked fine but we've not yet done a real comparison. I have a blue screen to use as well so next time we shoot for green I'll try to do a blue test.

In the mean time, people say green is better than blue so I use green. After all, time is limited on shoots and we all want to go home as soon as we can. I'm just skeptical as to why green is better than blue.

Rob Mack