A simple example of what’s the difference between premultiplied and straight footage: Here’s how a black pixel at 50% opacity is treated in two different cases:
- Premultiplied with white: in RGB the pixel is saved as grey (half-transparent black on white) and the Alpha-Channel is the same grey (fully opaque is white, fully transparent is black, so 50% opaque is 50% grey)
- Straight: RGB: any transparency is being ignored, the black pixel is saved as straight black. Alpha: 50% grey
Note: Premultiplied with black would coincidentally yield the same result as Straight in this case – but not if your image contains anything else than black and transparent blacks!
No matter what the RGB is, the Alpha is always the same.
So why would anyone ever want to use Premultiplied?
Well, 2 reasons:
- Some programs can divide the background-color out of the image and display a proper result (check the Footage Interpretation dialogue in After Effects for example), it’s only a problem when a program interprets Premutliplied footage as Straight…
- because premultiplied footage is easier to look at and doesn’t look broken to anyone who isn’t “in the know”…
