Re: proposal: background-image-alpha

> 
> What does "linear encoding for the mask" mean here?

After pre-converting the actual image to gamma 1.0, you multiply by the
values in the mask without doing any additional linearisation.  (You
then post convert the image to sRGB gamma.)

More precisely, you first normalise the mask and multiply the foregroound
by the mask value and the background by 1 - the mask value.

If your mask was in sRGB, but with zero chrominance, you would have to
cancel the sRGB gamma on both the mask and the image before combining.

Received on Tuesday, 9 May 2006 21:04:53 UTC