15843 2012-02-02 02:03:50 +0000 Handling color transition to/from transparent 2012-02-17 21:57:31 +0000 1 1 1 Unclassified CSS Transitions unspecified PC All RESOLVED DUPLICATE 14624 P2 normal --- 1 sylvaing dino ayg cmarrin dbaron eoconnor smfr dave.null oldest_to_newest 63573 0 sylvaing 2012-02-02 02:03:50 +0000 As CSS defines transparent to be 'transparent black' and we have agreed to transition instead to the transparent version of the other end of the transition - I believe Firefox does this - we need to capture this in the spec. I thought it was there but couldn't find it. 64236 1 ayg 2012-02-17 16:09:16 +0000 We should just define that color transitions happen in premultiplied space. Test case: data:text/html,<!doctype html> <body style="width:100px;color:green; transition:color 5s linear"> The quick brown fox jumped over the lazy dog <script> setTimeout(function(){document.body.style.color = 'rgba(7,250,9,0)'},0) setTimeout(function(){document.documentElement.textContent = getComputedStyle(document.body).color},2500) </script> Add a prefix to the 'transition' property as appropriate. Chrome 18 dev and Firefox trunk output "rgba(0, 128, 0, 0.5)" or something close to it, so they're using premultiplied space. IE10 Developer Preview and Opera Next 12.00 alpha output "rgba(3, 189, 4, 0.5)" or something close to it, so they're doing non-premultiplied space. The expected visual effect is always to do it in premultiplied space -- otherwise the transition can look quite weird. This shouldn't be a special case for "transparent". This is currently issue 6 in the spec. I suggest it be resolved in favor of the Gecko/WebKit behavior. 64238 2 smfr 2012-02-17 16:23:48 +0000 Agreed, although matching CSS gradients has been discussed in the past, and Mac WebKit (at least) can't do premultiplied gradients right now. 64255 3 dbaron 2012-02-17 19:15:16 +0000 Can we mark this a duplicate of bug 14624? 64264 4 sylvaing 2012-02-17 21:57:31 +0000 *** This bug has been marked as a duplicate of bug 14624 ***