- From: Matt Woodrow <mwoodrow@mozilla.com>
- Date: Thu, 22 Sep 2016 13:28:17 +1200
- To: Tien-Ren Chen <trchen@chromium.org>
- Cc: "Tab Atkins Jr." <jackalmage@gmail.com>, Rik Cabanier <cabanier@gmail.com>, /#!/JoePea <trusktr@gmail.com>, Chris Harrelson <chrishtr@google.com>, Simon Fraser <simon.fraser@apple.com>, "public-fx@w3.org" <public-fx@w3.org>
On 22/09/16 12:41 PM, Tien-Ren Chen wrote: > I don't think group opacity can be properly defined without > flattening. What if some planes in the group are non-opaque themselves > and later intersect with non-opaque planes outside of the group? > > Specifically, how would you define the expected behavior for this > particular example? http://jsbin.com/tepixi/ > Right, I totally agree with this. Opacity *does* require flattening, and if there are planes outsides the group that want to be sorted with planes within the group, there isn't a way to properly define behaviour. My point is that when we do the flattening for opacity, we should do so in the coordinate space of the root element of the preserve-3d context (by propagating ancestors transforms down through the opacity), since this results in better rendering for the cases where sorting across the opacity boundary isn't required (and there *is* a reasonable definition for the right behaviour). For cases like yours, there is no right behaviour, so we're just picking which fallback sucks the least. - Matt
Received on Thursday, 22 September 2016 01:28:55 UTC