1258132 - Inset box shadows doesn't transform:scale correctly

Status: RESOLVED DUPLICATE of bug 1250947

(Core :: Graphics, defect)

Description

[javi.gg]

Attachment: bug.jpg

User Agent: Mozilla/5.0 (Windows NT 5.1) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/49.0.2623.87 Safari/537.36

Steps to reproduce:

Create two identical divs sized in mm, apply the same inset box shadow with dimensions in mm, and transform:scale(0.5) the second


Actual results:

Second layer inset shadow renders incorrectly if it's inset, but renders correctly if not.


Expected results:

Inset shadows should be rescaled correctly as non-inset shadows do

Comment 1

[Tooru Fujisawa [:arai]]

might be related to bug 1188075 or bug 1227216 or similar bugs.

Just to be sure, can you post the HTML file used in the screenshot?

Comment 2

[javi.gg]

(In reply to Tooru Fujisawa [:arai] from comment #1)
> might be related to bug 1188075 or bug 1227216 or similar bugs.
> 
> Just to be sure, can you post the HTML file used in the screenshot?

You have all the relevant styles and HTML in the screenshot :-? (right side, inspector).

P.S. Results also vary (to even worse :) when a border-radius is set

Comment 3

[javi.gg]

Attachment: testcase.html

Comment 4

[javi.gg]

FF45 does it slightly better (not perfect), but again it goes nuts when setting border-radius on the test case.

Outset shadows perform ok on both (apparently).

Comment 5

[javi.gg]

Attachment: It's looks even worse when a border-radius is set

Comment 6

[javi.gg]

Attachment: bug_borderradius.jpg

Comment 7

[Tooru Fujisawa [:arai]]

Thank you for the testcases :)

The no radius case is partially fixed in bug 1227216 (Firefox 45), but not fully fixed, as there is small gray area between the shadow and the border.
  https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=c8dc34a351704525de17dd5d50d882a4c2339b5e&tochange=46c095a4311b63b5b71a9384169f1b109806d8c0

Also, the border-radius case shows the same issue on Nightly, it's possibly a same bug as bug 1250947.

Comment 8

[Tooru Fujisawa [:arai]]

Attachment: The rendering result with the patch in bug 1250947, on OSX.

Here's the comparison of the rendering result between Nightly (2016-03-17), and m-c b3006e4e09af + the patch in bug 1250947.
it looks better with the patch, the border-radius cases still have very small gray area between the border and the shadow tho, it might be a different issue related to smoothing or something.

mchang, can this bug be marked as DUPE of bug 1250947?

Comment 9

[javi.gg]

(In reply to Tooru Fujisawa [:arai] from comment #8)
> Created attachment 8732564 [details]
> The rendering result with the patch in bug 1250947, on OSX.
> 
> Here's the comparison of the rendering result between Nightly (2016-03-17),
> and m-c b3006e4e09af + the patch in bug 1250947.
> it looks better with the patch, the border-radius cases still have very
> small gray area between the border and the shadow tho, it might be a
> different issue related to smoothing or something.
> 
> mchang, can this bug be marked as DUPE of bug 1250947?

Agree to both replies :) WAY better with the patch :)

What is the estimated date for the mainstream release? Is there any way to get the patch and at least apply to my home environment? (a bit critical for a product development, relying exactly on inset shadows and rescaling (scale model custom license plate decals, printing right from CSS and a bit of html ;) )

I can workarround it more or less by manually doing the math for the shadow parameters for each scale class (1/12, 1/20, 1/24 and 1/43) and getting rid of the radius, but a little bit annoying... and license plates DO have a corner radius ;)

One billion thanks for your time, commitment, knowledge, and on and on and on ;)

Comment 10

[Mason Chang [Inactive] [:mchang]]

(In reply to Tooru Fujisawa [:arai] from comment #8)
> Created attachment 8732564 [details]
> The rendering result with the patch in bug 1250947, on OSX.
> 
> Here's the comparison of the rendering result between Nightly (2016-03-17),
> and m-c b3006e4e09af + the patch in bug 1250947.
> it looks better with the patch, the border-radius cases still have very
> small gray area between the border and the shadow tho, it might be a
> different issue related to smoothing or something.
> 
> mchang, can this bug be marked as DUPE of bug 1250947?

Thanks for your investigation. Yup, marking as a dupe of bug 1250947.