Re: SVG Compositing

RESEND TO THE GROUP...

On Mon, Aug 6, 2012 at 10:50 PM, Rik Cabanier <cabanier@gmail.com> wrote:

> Hi Bob,
>
> can you sent this to the group? The mailing list is archived and it is a
> valuable reference tool for us.
> Also, can you include Nikos in your emails?
>
> On Mon, Aug 6, 2012 at 4:08 AM, Bob Holmes <rangsynth@gmail.com> wrote:
>
>> One more note...
>>
>> 1) The formulas for composite modes... It is unclear if the "as" or
>> Alpha of source or even alpha of dest is from the pixel befre blending
>> or after blending.
>>
>> In other words the blending function computes a new value. How does
>> the alpha of this value relate to the composite function.
>>
>
> alpha is not affected by blending.
> Do you think we should call that out?
>
*I definitely think that this needs to be mentioned more. Are you saying
that for all the blending functions  that only RGB is used. So for MULTIPLY
mode it would only be r*r, g*g and b*b and leave the alpha alone?*
*
*
*If that is so then it is a huge distinction that should be at least given
its own small section. The whole time I have been grappling with how the
blending function is changing the alpha and then trying to work out what to
do with a MULTIPLY when drawing onto RGBA(0,0,0,0) which would always
produce 0 for example.*

>
>

>
>>
>> 2) The notation of cs to describe premultipled and Cs to describe
>> unpremul is very confusing!
>>
>
> That is fairly standard (ie http://en.wikipedia.org/wiki/Alpha_compositing
> )
> I think there will be pushback if we invent our own.
>
> Thanks. I think maybe in the spec it will help if there can be hyperlinks
to the notation legend. Like when you see a graph the legend is just there,
but for the formulas I missed that part at first and it was confusing.