[whatwg] Canvas Proposal: aliasClipping property

Charles Pritchard chuck at jumis.com
Thu Oct 8 15:25:57 PDT 2009


Robert O'Callahan wrote:
> On Fri, Oct 9, 2009 at 10:51 AM, Charles Pritchard <chuck at jumis.com 
> <mailto:chuck at jumis.com>> wrote:
>
>     Here is a good example of the conflict between Google and Mozilla
>     over clip():
>     http://www.imperialviolet.org/2009/09/02/anti-aliased-clipping.html
>
>     There is another conflict, with the composition modes between the
>     two, with Google
>     again taking another principled stand. I'm sorry I don't
>     supporting links for that,
>     but it effects many of the composite operations (at least four of
>     them).
>
>
> No, it is entirely about antialiasing. Coverage-based antialiasing 
> causes problems whenever you draw two objects with adjacent edges that 
> aren't aligned to pixel boundaries. It doesn't matter whether those 
> boundaries are induced by clipping or just by filling a path. The post 
> you linked to even refers to this:
It also refers to Chrome's 1-bit clipping path.

I apologize for not having a better reference -- several of the 
composite results
on the following resource have quite different results on Chrome.

https://developer.mozilla.org/en/Canvas_tutorial/Compositing

Should I perhaps take a different term, and rename this as a masking issue?
Would that be more accurate?

Chrome has taken a firm stance on how masking should be implemented,
Firefox has taken another view. This seems to be something that could be
mediated if the standard gives room for both of them, through a simple 
boolean toggle.

If I it were called "strictMasking" or something like that, would you 
support it?
I want to restrict its effects to the current divide in clip() and 
globalCompositeOperation
because that's where the trouble is. I'm not trying to add a new 
feature, just trying
to bring two implementations to the same standard.


-Charles



More information about the whatwg mailing list