Image formats for gnus-(x-)face-from-file

Image formats for gnus-(x-)face-from-file

[Reiner Steib]

[-- Attachment #1: Type: text/plain, Size: 645 bytes --]

Hi,

the following patch allows for use different input image formats by
using convert(1) to generate Face and X-Face headers.
Currently `gnus-x-face-from-file' requires GIF and
`gnus-face-from-file' requires JPEG.

For me, the current `gnus-x-face-from-file' command doesn't even work
(and I saw similar reports on Google):

$ giftopnm face.gif | ppmnorm | pnmscale -width 48 -height 48 | ppmtopgm |\
  pgmtopbm | pbmtoxbm | compface
giftopnm: Reading Image Sequence 0
ppmnorm: remapping 65..233 to 0..255
compface: (warning) <stdin>: excess data ignored
compface: (warning) <stdin>: excess data ignored
$


Is it okay to change the defaults?


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: rs-gnus-fun-face-w-convert.patch --]
[-- Type: text/x-patch, Size: 2861 bytes --]

--- gnus-fun.el	29 Aug 2005 16:37:45 +0200	7.9
+++ gnus-fun.el	13 Oct 2005 21:56:20 +0200	
@@ -45,21 +45,37 @@
   :group 'gnus-fun
   :type 'string)
 
-(defcustom gnus-convert-image-to-x-face-command "giftopnm %s | ppmnorm | pnmscale -width 48 -height 48 | ppmtopgm | pgmtopbm | pbmtoxbm | compface"
+(defcustom gnus-convert-image-to-x-face-command
+  "convert -resize 48x48 %s xbm:- | xbm2xface.pl"
   "Command for converting an image to an X-Face.
+The command must take a image filename (use \"%s\") as input.
+The output must be the Face header data on stdout in PNG format.
+
 By default it takes a GIF filename and output the X-Face header data
 on stdout."
   :version "22.1"
   :group 'gnus-fun
-  :type 'string)
+  :type '(choice (const :tag "giftopnm, netpbm (GIF input only)"
+			"giftopnm %s | ppmnorm | pnmscale -width 48 -height 48 | ppmtopgm | pgmtopbm | pbmtoxbm | compface")
+		 (const :tag "convert"
+			"convert -resize 48x48 %s xbm:- | xbm2xface.pl")
+		 (string)))
 
-(defcustom gnus-convert-image-to-face-command "djpeg %s | ppmnorm | pnmscale -width 48 -height 48 | ppmquant %d | pnmtopng"
+(defcustom gnus-convert-image-to-face-command
+  "convert -resize 48x48 %s -colors %d png:-"
   "Command for converting an image to a Face.
-By default it takes a JPEG filename and output the Face header data
-on stdout."
+
+The command must take an image filename (first format
+argument\"%s\") and the number of colors (second format argument:
+\"%d\") as input.  The output must be the Face header data on
+stdout in PNG format."
   :version "22.1"
   :group 'gnus-fun
-  :type 'string)
+  :type '(choice (const :tag "djpeg, netpbm (JPG input only)"
+			"djpeg %s | ppmnorm | pnmscale -width 48 -height 48 | ppmquant %d | pnmtopng")
+		 (const :tag "convert"
+			"convert -resize 48x48 %s -colors %d png:-")
+		 (string)))
 
 (defcustom gnus-face-properties-alist (if (featurep 'xemacs)
 					  '((xface . (:face gnus-x-face)))
@@ -125,8 +141,11 @@
 
 ;;;###autoload
 (defun gnus-x-face-from-file (file)
-  "Insert an X-Face header based on an image file."
-  (interactive "fImage file name (by default GIF): ")
+  "Insert an X-Face header based on an image file.
+
+Depending on `gnus-convert-image-to-x-face-command' it may accept
+different input formats."
+  (interactive "fImage file name: ")
   (when (file-exists-p file)
     (gnus-shell-command-to-string
      (format gnus-convert-image-to-x-face-command
@@ -134,8 +153,11 @@
 
 ;;;###autoload
 (defun gnus-face-from-file (file)
-  "Return a Face header based on an image file."
-  (interactive "fImage file name (by default JPEG): ")
+  "Return a Face header based on an image file.
+
+Depending on `gnus-convert-image-to-face-command' it may accept
+different input formats."
+  (interactive "fImage file name: ")
   (when (file-exists-p file)
     (let ((done nil)
 	  (attempt "")

[-- Attachment #3: Type: text/plain, Size: 114 bytes --]


Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: Image formats for gnus-(x-)face-from-file

[James Cloos]

I don't know how relevant it is for face headers, but I just noticed
that convert -resize 48x48 does not guarantee a 48x48 bit image, but
instead maintains the original's aspect ratio and sets the resulting
image's largest dimension to 48 pixels.  My test image came out as a
36x48 pixel xbm or png.

-JimC
-- 
James H. Cloos, Jr. <cloos@jhcloos.com>

Re: Image formats for gnus-(x-)face-from-file

[Reiner Steib]

On Fri, Oct 14 2005, James Cloos wrote:

> I don't know how relevant it is for face headers, but I just noticed
> that convert -resize 48x48 does not guarantee a 48x48 bit image, but
> instead maintains the original's aspect ratio and sets the resulting
> image's largest dimension to 48 pixels.  My test image came out as a
> 36x48 pixel xbm or png.

So we should use "convert -scale 48x48!"  instead of "convert -resize
48x48"?  Thanks for the hint.

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

face quantization (was: Image formats for gnus-(x-)face-from-file)

[Didier Verna]

Reiner Steib <reinersteib+gmane@imap.cc> wrote:

> On Fri, Oct 14 2005, James Cloos wrote:
>
>> I don't know how relevant it is for face headers, but I just noticed
>> that convert -resize 48x48 does not guarantee a 48x48 bit image, but
>> instead maintains the original's aspect ratio and sets the resulting
>> image's largest dimension to 48 pixels.  My test image came out as a
>> 36x48 pixel xbm or png.
>
> So we should use "convert -scale 48x48!"  instead of "convert -resize
> 48x48"?  Thanks for the hint.

        I don't know why, but recent modifications on the face code broke my
faces. They seem to have much less colors than before, while being of the same
size (in the header I mean). Has the quantization algorithm been modified
recently ?

-- 
Didier Verna, didier@lrde.epita.fr, http://www.lrde.epita.fr/~didier

EPITA / LRDE, 14-16 rue Voltaire   Tel.+33 (1) 44 08 01 85
94276 Le Kremlin-Bicêtre, France   Fax.+33 (1) 53 14 59 22   didier@xemacs.org

Re: face quantization

[Reiner Steib]

On Thu, Oct 27 2005, Didier Verna wrote:

>         I don't know why, but recent modifications on the face code
> broke my faces. They seem to have much less colors than before,
> while being of the same size (in the header I mean). Has the
> quantization algorithm been modified recently ?

The algorithm in `gnus-face-from-file' itself hasn't changed.
But the default of `gnus-convert-image-to-face-command' is
  "convert -scale 48x48! %s -colors %d png:-"
instead of
  "djpeg %s | ppmnorm | pnmscale -width 48 -height 48 | ppmquant %d | pnmtopng".

It's quite possible that the convert differs from the later.  Could
you investigate?

My intention was to allow more input formats (for both, X-Face and
Face [1]).  But if convert is not doing a good job, we could change
the default back to use djpeg and the netpbm tools.

Bye, Reiner.

[1] Cf. your comment in `gnus.texi':

| 6.528 (didier 18-Jun-03):
| @c #### FIXME: faces and x-faces'implementations should really be harmonized.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: face quantization

[Didier Verna]

Reiner Steib <reinersteib+gmane@imap.cc> wrote:

> The algorithm in `gnus-face-from-file' itself hasn't changed.
> But the default of `gnus-convert-image-to-face-command' is
>   "convert -scale 48x48! %s -colors %d png:-"
> instead of
>   "djpeg %s | ppmnorm | pnmscale -width 48 -height 48 | ppmquant %d | pnmtopng".
>
> It's quite possible that the convert differs from the later.  Could
> you investigate?

        After some testing, it appears that convert does a better job (in the
sense that the result is closer to the original image), although I tend to
prefer the former solution because the results are brighter and a bit
"artificial", kinda special-effect :-)

        I also noticed that given my original jpg faces that are already at a
size of 48x48, the average quantization I get is 4 to 6 which is quite poor.
Maybe the size limit of 726 has something to do with it. As it changed in the
last few months (I cound't find a trace of this in the ChangeLog) ?


        Finally, it might be worth considering that after quantization to 4 or
6 colors, the resulting PNG images are *bigger* than my original JPEG ones.
So why not using JPG in the headers instead ?

-- 
Didier Verna, didier@lrde.epita.fr, http://www.lrde.epita.fr/~didier

EPITA / LRDE, 14-16 rue Voltaire   Tel.+33 (1) 44 08 01 85
94276 Le Kremlin-Bicêtre, France   Fax.+33 (1) 53 14 59 22   didier@xemacs.org

Re: face quantization

[Reiner Steib]

On Fri, Oct 28 2005, Didier Verna wrote:

>         I also noticed that given my original jpg faces that are already at a
> size of 48x48, the average quantization I get is 4 to 6 which is quite poor.
> Maybe the size limit of 726 has something to do with it. As it changed in the
> last few months (I cound't find a trace of this in the ChangeLog) ?

It hasn't changed since...

,----[ C-x v g / vc-annotate ]
| revision 6.45
| date: 2003/04/21 11:42:07;  author: kaig;  state: Exp;  lines: +3 -3
| (gnus-face-from-file, gnus-convert-png-to-face):
| Max length of header is 726, not 740.  From Gaute B Strokkenes
| <gs234@cam.ac.uk>.
`----

>         Finally, it might be worth considering that after
> quantization to 4 or 6 colors, 

BTW, would it make sense to decrease by one instead of two at least if
quant becomes small?

--8<---------------cut here---------------start------------->8---
--- gnus-fun.el	14 Oct 2005 18:51:50 +0200	7.10
+++ gnus-fun.el	28 Oct 2005 12:50:42 +0200	
@@ -172,7 +172,7 @@
 			 quant))))
 	(if (> (length attempt) 726)
 	    (progn
-	      (setq quant (- quant 2))
+	      (setq quant (- quant (if (< quant 10) 1 2)))
 	      (gnus-message 9 "Length %d; trying quant %d"
 			    (length attempt) quant))
 	  (setq done t)))
--8<---------------cut here---------------end--------------->8---

> the resulting PNG images are *bigger* than my original JPEG ones.
> So why not using JPG in the headers instead ?

Ask Lars; his specification says PNG.  :-)

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: face quantization

[Didier Verna]

Reiner Steib <reinersteib+gmane@imap.cc> wrote:

>>         Finally, it might be worth considering that after quantization to 4
>> or 6 colors, the resulting PNG images are *bigger* than my original JPEG
>> ones. So why not using JPG in the headers instead ?
>
> Ask Lars; his specification says PNG.  :-)

        OK. Lars ? :-)

-- 
Didier Verna, didier@lrde.epita.fr, http://www.lrde.epita.fr/~didier

EPITA / LRDE, 14-16 rue Voltaire   Tel.+33 (1) 44 08 01 85
94276 Le Kremlin-Bicêtre, France   Fax.+33 (1) 53 14 59 22   didier@xemacs.org

Re: face quantization

[Lars Magne Ingebrigtsen]

Didier Verna <didier@lrde.epita.fr> writes:

>>>         Finally, it might be worth considering that after quantization to 4
>>> or 6 colors, the resulting PNG images are *bigger* than my original JPEG
>>> ones. So why not using JPG in the headers instead ?
>>
>> Ask Lars; his specification says PNG.  :-)
>
>         OK. Lars ? :-)

When "designing" this, I did quite a few tests with jpeg/png 48x48
images constrained to 778 bytes, and the jpeg ones always looked much,
much worse than the PNG ones.

You might find examples of the opposite, of course.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen