I re-read the following:
Y = 0.299R + 0.587G + 0.114B
A light bulb went on, and I realised I'd been completely wrong in the way I was scaling my Red, Green and Blue images before combining them. This also meant that the Gaussian Blur step was not needed.
What I needed to do to with the colour channel images I made from my initial 720 x 576 images was to horizontally scale them in The GIMP by following amounts:
- Red to 720 x 0.299 = 215 pixels
- Green to 720 x 0.587 = 423 pixels
- Blue to 720 x 0.114 = 82 pixels
Here's the logo I created for my friend Greg Taylor's company before:
Before image. Can you spot Emily Moore?
And here it is after using this revised method:
Now with PAL - click to enlarge the images
I'm sure I'll come back to this again, but I'm finding it interesting.
5 comments:
Theoretically, these figures can also be used to simulate telly transmissions from the now-defunct (in the U.S.) NTSC system, as they used the same mathematical figures for color.
I have never seen NTSC originated pictures on a domestic NTSC television - I would have loved to have seen how the system looked to compare it to PAL.
In addition, the horizontal pixel rate on the 'Y' matrix differs among NTSC, PAL and SECAM. They are as follows:
NTSC - 330 pixels
PAL - 425 pixels
SECAM - 465 pixels
Many thanks for that wbhist - I'm planning on altering my plug-ins shortly, so I'll add these values, and maybe add NTSC and SECAM as well.
In addition, if doing simulated NTSC screencaps, the recommended width and height is 656 x 486, with the "clean" aperture being 648 and four "blank" horizontal pixels on each edge (eight total). Of course, for DVD's and DV's the top and bottom are shaved off to 480, but still . . . 486 is "D1" standard.
Post a Comment