Re: [css3-images] `image-resolution: span` and transformations.

Le 29/07/2013 17:33, Tab Atkins Jr. a écrit :
> On Mon, Jul 29, 2013 at 4:10 AM, Simon Sapin <simon.sapin@exyr.org> wrote:
>> Le 26/07/2013 18:41, Tab Atkins Jr. a écrit :
>>> On Fri, Jul 26, 2013 at 10:21 AM, Simon Sapin<simon.sapin@exyr.org>
>>> wrote:
>>>>
>>>> Le 26/07/2013 17:56, Tab Atkins Jr. a écrit :
>>>>>
>>>>> I don't think it should.  For one, this would mean that the intrinsic
>>>>> size of an image changes as you transform it, which is clearly not a
>>>>> good result.
>>>>
>>>>
>>>> Your first point also applies to user zoom, especially with mobile-style
>>>> panning zoom. What does "snap" mean in this context?
>>>
>>> No, mobile pinch-zoom is a distinct type of zoom.  We need to
>>> formalize these concepts within CSS, as they're being formalized in
>>> the back-end ad-hocly right now.
>>>
>>> The relevant type of zoom is the one that changes the viewport size.
>>> This changes a bunch of the layout, so it's okay for images to have a
>>> different intrinsic size.
>>
>> Ok. So, to sum up:
>>
>> * Transforms should not affect 'snap'. This makes 'snap' useless on
>> transformed images, but that seems better than the alternative.
>>
>> * "Desktop-type" zoom that changes the size (in CSS units) of the initial
>> containing block should affect 'snap'. That’s fine because layout probably
>> changes anyway.
>>
>> * "Mobile-type" zoom that does not change the size of the ICB should not
>> affect 'snap'. IMO this makes 'snap' completely useless if the user is
>> expected change the zoom level a lot. I’d rather have UAs ignore it in such
>> cases than apply it in a way that only makes sense at one precise zoom
>> level.
>
> If you're designing a mobile site and care about the resolution of
> your images, you should be setting the zoom level to 1 anyway.  Users
> are *not* generally "expected to change the zoom level a lot" - when
> that's happening, it's because they're using a site designed solely
> for desktop without any thought given to mobile.
>
> Additionally, as I stated, these constraints are exactly the same as
> what we want to expose for <canvas>.


My point is, "number of device pixels" (and its relationship to CSS 
units) needs a much more precise definition than what’s currently in 
css-images-3, and we may want to tweak that definition with respect to 
transforms, zoom in various situations, etc.

I don’t have a strong opinion on the details of this definition, as long 
as it’s well-defined. Do you have a wording to suggest?

-- 
Simon Sapin

Received on Monday, 29 July 2013 16:54:48 UTC