- From: Alan Stearns <stearns@adobe.com>
- Date: Fri, 18 Oct 2013 13:12:39 -0700
- To: fantasai <fantasai.lists@inkedblade.net>, "www-style@w3.org" <www-style@w3.org>
On 10/18/13 11:56 AM, "fantasai" <fantasai.lists@inkedblade.net> wrote: >On 10/18/2013 11:52 AM, fantasai wrote: >> On 10/17/2013 07:41 PM, Alan Stearns wrote: >>> fantasai wrote: >>>> >>>> Okay, now let's consider an element that contains text, but has a >>>>shape >>>> derived from an image. This means that the size of the image and what >>>> it's box would be is potentially very different from the size of the >>>> box resulting from sizing the text. That size is therefore mismatched, >>>> though well-defined. What is its position with respect to the actual >>>> box? >>> >>> I would expect the top-left corner of the image would be in the >>>top-left >>> corner of the content box. What would you suggest? >> >> I would suggest centering the image. Any time we pick a side, it's gives >> us a bias. That has two problems: it's non-symmetrical--and most designs >> are better off if we're being symmetrical--and it's i18n-unfriendly. > >Btw, we might actually want to key off object-fit/object-position instead, >taking the used content box size as the image's content box for this >purpose. >This not only guarantees that the images match, but also let's the author >play around with settings in the case that the element being shaped is >not a replaced image. Since object-fit and object-position are part of the 'rules that would apply' I agree it makes sense to use them. And object-position's initial value makes the image centered, so we get your desired result. You suggested this wording on IRC today: --- The image is sized and positioned as if it were a replaced element with a specified width and height of the element's used content-box size. --- This works for the case I'm concerned about (an <img> float where we're getting the shape from the displayed image). And it gives a clear result for all other cases where the image isn't actually the displayed content. If there's a mismatch, then object-position will center the image in its initial state, but you can fiddle with the image placement with different object-position values. I've edited in the sentence above. Thanks, Alan
Received on Friday, 18 October 2013 20:13:05 UTC