Difference between revisions of "Film Projects"

From ivc wiki
Jump to navigationJump to search
Line 4: Line 4:


=== YUV color space ===
=== YUV color space ===
[[Image:YCbCr.jpg|right]]
[[Image:YCbCr.jpg|right|YUV channels, full picture, Y, U, and V]]


* Y: [http://en.wikipedia.org/wiki/Luma_(video) Luma] - Brightness black-and-white signal
* Y: [http://en.wikipedia.org/wiki/Luma_(video) Luma] - Brightness black-and-white signal

Revision as of 16:06, 6 April 2008

Blue or Greenscreen

Aka. chroma key, and color keying.

YUV color space

  • Y: Luma - Brightness black-and-white signal
  • UV: Chroma (blue and red components) - Color signal

YUV is not an absolute color space, it's a way of encoding RGB information [1].

Chroma resolution resampling

  • 4:4:4 (Chroma:Cr:Cb - Y:U:V) - full resolution / high bandwidth
  • 4:2:0 (Chroma:Cr:Cb - Y:U:V) - less color resolution / lower bandwidth
  • 4:1:1 (Chroma:Cr:Cb - Y:U:V) - less color resolution / lower bandwidth

MacBreak part 2 explains and shows how chroma subsampling works.

Because of storage and transmission limitations, there is always a desire to reduce (or compress) the signal. [2]

The green information is usually preserved in the Y channel [3]. Which makes a green background perfect because of the extra channel resolution on 4:2:0 and 4:1:1 based cameras. In addition green is not a color usually found on human beings, making it easy to key.

Quotes

If you can plan your DV shot to have mostly dark subjects (dark hair & clothing) in front 
of a green screen (or bright/blond subjects in front of blue), you can use the color difference 
key to pull a quick core matte, then turn to a luma-based solution to bring out the edge details.
With DV and DVCam, you won't have as much color information per pixel due to compression, so 
good mattes are harder to pull. It ends up looking pretty aliased. Both NTSC and Pal DV formats 
have pretty low sampling (4:1:1 for NTSC and 4:2:0 for Pal).
It was all pixelated and jagged edges. How do u get around this?
The technique that I've heard used most often is that you blur the UV channels (the chromanance 
part of the YUV) then mix that back in with the Y (luminance) and your ready to pull a matte. Do 
just enough blur so the pixelated jagged edges soften. Use that only for pulling the matte. Once 
you've got your matte, go back and use the original footage for applying the matte.

Programs / Plug-ins

  • dvMatte Pro - Final Cut Pro, OSX
  • Keylight - After Effects
  • Shake - OS X only
  • Final Cut Pro - OS X only
  • Boris Key filters - After Effects, Final Cut Pro, etc

Lightning