[sdl.web@gmail.com: Re: problem with transparent PNG image display]

[sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

Would someone please debug this, then ack?

------- Start of forwarded message -------
X-Spam-Status: No, score=-0.0 required=5.0 tests=SPF_HELO_PASS,SPF_PASS,
	UNPARSEABLE_RELAY autolearn=failed version=3.1.0
To: emacs-pretest-bug@gnu.org
From: Leo <sdl.web@gmail.com>
Date: Sun, 17 Jun 2007 10:14:40 +0100
Mime-Version: 1.0
Content-Type: multipart/mixed; boundary="=-=-="
Subject: Re: problem with transparent PNG image display

- --=-=-=

This follows from article: <m2ac0rogco.fsf@gmail.com> in pretest-bugs
list.

For example the attached icon has transparent background but shows a
white background in Emacs.

- -- 
Leo <sdl.web AT gmail.com>                         (GPG Key: 9283AA3F)

- --=-=-=
Content-Type: image/png
Content-Disposition: attachment; filename=system-users.png
Content-Transfer-Encoding: base64

iVBORw0KGgoAAAANSUhEUgAAACAAAAAgCAYAAABzenr0AAAABmJLR0QA/wD/AP+gvaeTAAAAB3RJ
TUUH1gsKFTktQWg0swAAB8FJREFUWIXlV1tsFNcZ/mbO3HZ2d/bitfGFi7HXDuZaSAqikDYIVUFC
abi4BEopD3nrG1JKGlAqiwhh16h+qPrQVkrVh7RVUiyoooBKKJcQATFp7OILF8uG2JBiK7az6/XO
5Vz6MLtjY7kEqvapR/o1u2eOzvnO93/n+88A/+9N+roBbW1tcY/n9yiKuplSWgYAiqKMUOqdU+XQ
Hw8cODDxPwPQ0nJ0k6Krb659bq1aX7dEsawYACCbzeDOndv0ascVjzreW6+/fvj8fx3A8ePNG81I
5MiObTtD4XAEsgzIsgJZlsE5B8CRzU7iz+3v5acmJ3/22ms/vfyfACBzdTY1NZm6qf+qcfuuiGka
EIKDEAJFUaGqKoQQ8DwPiqIgXVun9tzsXrd+3Yb2CxcueE8LQJmrMxrVd6/5xhrTMFQ4jgNV1cC5
AGMMQhSfHJ7ngjGmJuMlSwWXBn9+/FgcAFRVHaLUe9dzRNuhQ4dGHwdgzhQcb2t9r3F743xV9SlX
VRWKooAQBYAEITg4p+jp6U1+cv1aeu031yFdWyc/opH+297Va1dc5rH9Bw++ceLfAZDn6uTUm2ea
JhjzwBgDpR4opfA8txAebt68leju7a7d88peedXK1XIkEvEnlGVEoxbWrF6j7tu7Pxy2Ir9vbW1+
6akAAIAQHJRScE5BKSss7MDzXORyWdLx6bXal7Z+j+i6Ds9zAQCEEKiqCkmSQCmDYRjYua0xrGjy
71paWqJPDEAiykg2mwUAuK4X7Np1GTzPQ1dXZ8XKFStlXdfgug6EEOCcgzH/Pec80Iiu61i9+rko
IeLAEwOwc5PnBu8OUE0zwJgLSj24rs+A6zoYfnA/tWhhjew4Njjn4JyBMQpKfbCU+qljjMF1HdRU
12iyQvY9MYB79x6c6ezqkjzqQtdD8DwPjLmBBnK5SS0cNgv9FJTSQCM+UF8zPiMuTDMMSumCJwJw
tPXo8vol6V/PKyuTTp85DSGAWCwOWdYKu+UAUKCcBiKdFui0UIviFYLPtTaAWT7Q3Hxklabpbbt2
7g5JkoSOjg60n2rHyuWrsLi6BslkCgBgGKY3OZnVJQlBzjnnkCQZkoRAE36fhGw2C5mQ4ccCaDre
lNIVo3VX456QZVlgzMXSpQ2Ix2MYGOxHb18PHMfxd8/4lwODAxUNSxqkyclswRcYJKlIqADnApwz
mGYEfbf6XGcq/6fHAoip4SMvfndL2LIicJw8FEVBeXk5TNNESUkK+XwejDEAwMTE2Og/bnRVpNN1
MIwQpqZykGUZkuT7mhD+MQ6FTHjUQ1fXZ9LIyNhv5gIgA0BLW0tDLJasW7CgWs7n88EuAIFYLI7a
2hrU16dRW1sDI2REb9+5tawsVYGZGiFECWgnhMCyYhACOH3mNEpLSqWFi6ouHzt2rGZOABrkPevX
rQ8Xz7TjuODc34kPBAiFQhgfH4te/vjikp07dimbN2+WKsur0H7yBHr7+iDLBMlkCslkCrJM0Hez
D+0nT2Dh/EXYsmWr8vLL2+cbpnK1tfWtxTMBSADwi7aWD/b/6NWE6/pOJ0kEpmmAEKVALcHkVFY9
ceLdVTu2fV+xLAue52J0dBQPHnyBgcF+ZCezgUZ0XYcVjSFdW4+qqkokEgmoqoaR0Yfi1Kn2oa8m
cg1NTU1TgQY8yixN02DbeQwNfY50uh5CIDhykgR8ePZs3aYXNpNoNALbzoMQBeXlFTDNcKARSikA
QNNUmKaJRCKBcDgMxhgcJ4/SVKn0/PMvlF+8dP6XAF4NUkBk2WWMIpOZgKoaBVVPx/D9oQjAw4sW
VkuzNRKP+xqpq0sjna5FOp1GOp1GdfViWJYVUC0EYNt5PFP/jBZPJHYVUyEDgKyQz8fGx+A4Lior
K8CYf6w456CUobv7RsWzz64ljjOtEV/pAkJMayQWiyEWi8EwDAAieO+P8b3BtvP49sbvmEQ1DgcA
bNtr/6Sjw06lSoOJGePBxWNkdMSaVzYPruvCtu2g2HAuAraK3l+MIntC8ABENpuF57lIlaRkVVV3
BADe/u3bFwbvDnw5fH9YWJYFRfFvakUGbNtWNE0D5wzDw0PQNLUw6fTifkHyi1KxQD0Kwv+dyfhV
1rJi4ebm5pgCAP39/fQv779/VAhx5OHIw9iyhmVqMlkCVVUhyxIIIYIxKmUyX0FVtcJk05cpSRIQ
wrfhaSMSs7QkEA6b6OzsQmlpGVRFYVSjkaIT5q9cunKvp7PnwA/3/2BTz43uDYpGFgoBTQhBVFXN
j42Ph23bQWVlReF0FF1PFGqAeARAMecz64I/FsjlchifmFBoho7OvhMaADTMui0ffvONfSuWr2je
uGFjqDiJ7w9yYMFFG55WvQgWn8lEJpNFNpsVH12+ePXgTw59a/a1nAJwANgzY3Dg7p36JendyWRJ
Yn7VfKlI6Wylz4zZu5ZlAtMMQwiBv50/Z1/v+OzH3d3dg3N+F8xumUxGydn5Dl3TXszlcno8HidW
NIZQyEQoFIKuh2AYBnR9Oor/DSMEWSZwHAe9fb307Id/dTv/3tX0zjt/OA8g87XfhoUmAShLJBLz
9ux9pbGqsmqrppF6LmAIIeb8tpiZCkJITnAx+uCLf35w8aNLJzuvdw4AGH0aAMVmwtfJYxd9HB4A
HoBc4Yl/AfM5qxyUm8X5AAAAAElFTkSuQmCC
- --=-=-=
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
emacs-pretest-bug mailing list
emacs-pretest-bug@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-pretest-bug

- --=-=-=--
------- End of forwarded message -------

[sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

[I sent this message a few weeks ago but did not get a response.]

Would someone please debug this, then ack?

------- Start of forwarded message -------
X-Spam-Status: No, score=-0.0 required=5.0 tests=SPF_HELO_PASS,SPF_PASS,
	UNPARSEABLE_RELAY autolearn=failed version=3.1.0
To: emacs-pretest-bug@gnu.org
From: Leo <sdl.web@gmail.com>
Date: Sun, 17 Jun 2007 10:14:40 +0100
Mime-Version: 1.0
Content-Type: multipart/mixed; boundary="=-=-="
Subject: Re: problem with transparent PNG image display

- --=-=-=

This follows from article: <m2ac0rogco.fsf@gmail.com> in pretest-bugs
list.

For example the attached icon has transparent background but shows a
white background in Emacs.

- -- 
Leo <sdl.web AT gmail.com>                         (GPG Key: 9283AA3F)

- --=-=-=
Content-Type: image/png
Content-Disposition: attachment; filename=system-users.png
Content-Transfer-Encoding: base64
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- --=-=-=
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
emacs-pretest-bug mailing list
emacs-pretest-bug@gnu.org
http://lists.gnu.org/mailman/listinfo/emacs-pretest-bug

- --=-=-=--
------- End of forwarded message -------

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Jason Rumney]

Richard Stallman wrote:
> [I sent this message a few weeks ago but did not get a response.]
>
> Would someone please debug this, then ack?
>   
> For example the attached icon has transparent background but shows a
> white background in Emacs.
>   
Someone investigated before the release, and decided that the image
caching code needed overhauling, so this should be done in the trunk.

Because we cache images in their uncompressed and rendered form, we
cannot support full transparency for png images. So we use the
background color that is saved in the image itself to render the image.
When no background color is saved in the image, we render using the
background color of the face the image is first drawn with, so the
problem is not immediately apparent with those images, but if you draw
the same image object onto a different colored background, you will see
the same problem occur.

I don't know which platforms will let us cache image objects with
transparency, but the solution would be to do that, and draw the
background separately each time the image is drawn.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

    When no background color is saved in the image, we render using the
    background color of the face the image is first drawn with, so the
    problem is not immediately apparent with those images, but if you draw
    the same image object onto a different colored background, you will see
    the same problem occur.

The cache could record what background color was used for rendering.
Then, on displaying the image again, if the desired background is
different, that could cause a cache-mismatch, which would cause
re-rendering.

For efficiency, we should do that only for formats that allow
transparency.

Is it possible to tell easily whether a png image has any
transparency?  Then we could optimize the images that don't
have any transparency by not recording a background for it.

    I don't know which platforms will let us cache image objects with
    transparency, but the solution would be to do that, and draw the
    background separately each time the image is drawn.

There must be something I don't get.  What would stop us from caching
the actual image data, transparent or not?

Is the idea that we want to save the time to re-render the image?

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

[I sent this message a week ago but did not get a response.
Could we get the discussion moving again?]

    When no background color is saved in the image, we render using the
    background color of the face the image is first drawn with, so the
    problem is not immediately apparent with those images, but if you draw
    the same image object onto a different colored background, you will see
    the same problem occur.

The cache could record what background color was used for rendering.
Then, on displaying the image again, if the desired background is
different, that could cause a cache-mismatch, which would cause
re-rendering.

For efficiency, we should do that only for formats that allow
transparency.

Is it possible to tell easily whether a png image has any
transparency?  Then we could optimize the images that don't
have any transparency by not recording a background for it.

    I don't know which platforms will let us cache image objects with
    transparency, but the solution would be to do that, and draw the
    background separately each time the image is drawn.

There must be something I don't get.  What would stop us from caching
the actual image data, transparent or not?

Is the idea that we want to save the time to re-render the image?

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Jason Rumney]

Richard Stallman wrote:
> [I sent this message a week ago but did not get a response.
> Could we get the discussion moving again?]
>   

I don't know enough about the image code and PNG format to answer the
questions you raised. I replied to your previous message because I
remembered that someone had looked into this already, but noone else had
spoken up about it.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

[I sent this message a week ago but did not get a response.
Could we get the discussion moving again?]

    When no background color is saved in the image, we render using the
    background color of the face the image is first drawn with, so the
    problem is not immediately apparent with those images, but if you draw
    the same image object onto a different colored background, you will see
    the same problem occur.

The cache could record what background color was used for rendering.
Then, on displaying the image again, if the desired background is
different, that could cause a cache-mismatch, which would cause
re-rendering.

For efficiency, we should do that only for formats that allow
transparency.

Is it possible to tell easily whether a png image has any
transparency?  Then we could optimize the images that don't
have any transparency by not recording a background for it.

    I don't know which platforms will let us cache image objects with
    transparency, but the solution would be to do that, and draw the
    background separately each time the image is drawn.

There must be something I don't get.  What would stop us from caching
the actual image data, transparent or not?

Is the idea that we want to save the time to re-render the image?

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

    I don't know enough about the image code and PNG format to answer the
    questions you raised. I replied to your previous message because I
    remembered that someone had looked into this already, but noone else had
    spoken up about it.

OK, I will send it again, not addressed specifically to you.

We need SOMEBODY to look at this!

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Miles Bader]

Richard Stallman <rms@gnu.org> writes:
> The cache could record what background color was used for rendering.
> Then, on displaying the image again, if the desired background is
> different, that could cause a cache-mismatch, which would cause
> re-rendering.

I thought it already did this (in image.c:search_image_cache):

2007-06-13  YAMAMOTO Mitsuharu  <mituharu@math.s.chiba-u.ac.jp>

	* dispextern.h (struct image) [HAVE_WINDOW_SYSTEM]: New members
	frame_foreground and frame_background.

	* image.c (lookup_image): Save frame foreground and background colors.
	(search_image_cache): Check if saved and current frame colors match.

-Miles

-- 
Americans are broad-minded people.  They'll accept the fact that a person can
be an alcoholic, a dope fiend, a wife beater, and even a newspaperman, but if a
man doesn't drive, there is something wrong with him.  -- Art Buchwald

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Jason Rumney]

Miles Bader wrote:
> I thought it already did this (in image.c:search_image_cache):
>
> 2007-06-13  YAMAMOTO Mitsuharu  <mituharu@math.s.chiba-u.ac.jp>
>
> 	* dispextern.h (struct image) [HAVE_WINDOW_SYSTEM]: New members
> 	frame_foreground and frame_background.
>
> 	* image.c (lookup_image): Save frame foreground and background colors.
> 	(search_image_cache): Check if saved and current frame colors match.
>   

I think its the background on which the image is rendered that is
important, not the frame background. The toolbar has its own background
color, independent of the frame's. And that only covers caching, images
still need to be drawn with the appropriate background from the face
rather than the background color saved in the image file itself.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[YAMAMOTO Mitsuharu]

>>>>> On Wed, 01 Aug 2007 16:55:30 +0100, Jason Rumney <jasonr@gnu.org> said:

> Miles Bader wrote:
>> I thought it already did this (in image.c:search_image_cache):
>> 
>> 2007-06-13  YAMAMOTO Mitsuharu  <mituharu@math.s.chiba-u.ac.jp>
>> 
>> * dispextern.h (struct image) [HAVE_WINDOW_SYSTEM]: New members
>> frame_foreground and frame_background.
>> 
>> * image.c (lookup_image): Save frame foreground and background colors.
>> (search_image_cache): Check if saved and current frame colors match.
>> 

> I think its the background on which the image is rendered that is
> important, not the frame background. The toolbar has its own background
> color, independent of the frame's. And that only covers caching, images
> still need to be drawn with the appropriate background from the face
> rather than the background color saved in the image file itself.

FWIW, that change was just for fixing excessive image cache miss (and
performance degradation caused by needless image regeneration) in the
trunk, without affecting its original intention.

				     YAMAMOTO Mitsuharu
				mituharu@math.s.chiba-u.ac.jp

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Chong Yidong]

Richard Stallman <rms@gnu.org> writes:

>     When no background color is saved in the image, we render using the
>     background color of the face the image is first drawn with, so the
>     problem is not immediately apparent with those images, but if you draw
>     the same image object onto a different colored background, you will see
>     the same problem occur.
>
> The cache could record what background color was used for rendering.
> Then, on displaying the image again, if the desired background is
> different, that could cause a cache-mismatch, which would cause
> re-rendering.

There is code to handle this in the trunk, but not in the branch.

I just changed the trunk code further to ignore the background color
specified by the png image (now that Emacs can properly handle
transparency, it is more correct to always use the frame background).

If you want this bug resolved for Emacs 22, it will be necessary to
port both changes from the trunk.  I think it should be moderately
safe.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Jason Rumney]

Chong Yidong wrote:
> There is code to handle this in the trunk, but not in the branch.
>
> I just changed the trunk code further to ignore the background color
> specified by the png image (now that Emacs can properly handle
> transparency, it is more correct to always use the frame background).
>   

It should be the face background, not the frame background. My
understanding is that the existing code that checks the frame background
was to fix a different bug with image caching.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Chong Yidong]

Jason Rumney <jasonr@gnu.org> writes:

> Chong Yidong wrote:
>> There is code to handle this in the trunk, but not in the branch.
>>
>> I just changed the trunk code further to ignore the background color
>> specified by the png image (now that Emacs can properly handle
>> transparency, it is more correct to always use the frame background).
>
> It should be the face background, not the frame background. My
> understanding is that the existing code that checks the frame background
> was to fix a different bug with image caching.

Why should images be displayed with the face background?  That
behavior doesn't make sense, since images are not displayed with a
face.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Jason Rumney]

Chong Yidong wrote:
> Why should images be displayed with the face background?  That
> behavior doesn't make sense, since images are not displayed with a
> face.
>   

Yes they are. The tool-bar should be displayed with the tool-bar face
(likewise the mode-line and fringe).

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Richard Stallman]

    If you want this bug resolved for Emacs 22, it will be necessary to
    port both changes from the trunk.  I think it should be moderately
    safe.

Please do that -- this is a bug fix.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Chong Yidong]

Jason Rumney <jasonr@gnu.org> writes:

> Chong Yidong wrote:
>> Why should images be displayed with the face background?  That
>> behavior doesn't make sense, since images are not displayed with a
>> face.
>
> Yes they are. The tool-bar should be displayed with the tool-bar face
> (likewise the mode-line and fringe).

In the case of transparent PNG images, I think it's more reasonable to
expect the background color to be whatever is "underneath" the image.
I, for one, would be surprised if that is not the behavior.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Chong Yidong]

Richard Stallman <rms@gnu.org> writes:

>     If you want this bug resolved for Emacs 22, it will be necessary to
>     port both changes from the trunk.  I think it should be moderately
>     safe.
>
> Please do that -- this is a bug fix.

Done.

Re: [sdl.web@gmail.com: Re: problem with transparent PNG image display]

[Leo]

On 2007-07-23 19:07 +0100, Richard Stallman wrote:
> [I sent this message a few weeks ago but did not get a response.]
>
> Would someone please debug this, then ack?

This has indeed been fixed. Many thanks.

-- 
Leo <sdl.web AT gmail.com>                         (GPG Key: 9283AA3F)

         Gnus is one component of the Emacs operating system.