[OS X TeX] Re: Apple PDF viewer bugs on this shading

[Ross Moore]

Hi Quark (?)

On 01/12/2009, at 7:00 AM, Quark67 wrote:

> Hello, sorry for my english, I'm the original poster from the  
> french newsgroup which Alain Matthes speak.
>
> In your message http://www.tug.org/pipermail/macostex-archives/2009- 
> November/042073.html you have posted a screenshoot attachment and  
> you say "It looks fine for me, in both Preview and TeXshop.  " . I  
> suppose you have compiled this with tex+dvi in TexShop ?

No. It was compiled with pdftex as the engine.
When I do it again using TeX+dvi mode then the squares are more
prominent, as you say.


On the MacOS X TeX list the interest seemed to be in the results
obtained with different viewers. I seemed to be getting different
results from what others claimed, but they didn't post any images
to show what was the issue. Now that you have shown what is the
real problem, we can examine it in more detail.


Even with pdftex, the squares are visible when blowing-up to
~400% , and even noticeable at ~200%.
An attached image shows this, where the squares are just discernible
on the right, in Preview, but not on the left using Adobe Reader.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: withPdfTeX.png
Type: image/png
Size: 99806 bytes
Desc: not available
URL: <http://email.esm.psu.edu/pipermail/macosx-tex/attachments/20091201/217e8fff/attachment.png>
-------------- next part --------------


These were done using pdfTeX:
   This is pdfTeXk, Version 3.141592-1.40.3 (Web2C 7.5.6)


Using TeX+dvi+Ghostscript the squares are much more prominent.
Both PDF viewers seem to give similar results (as in the 2nd attached
image):

-------------- next part --------------
A non-text attachment was scrubbed...
Name: withDVI+GS.png
Type: image/png
Size: 41399 bytes
Desc: not available
URL: <http://email.esm.psu.edu/pipermail/macosx-tex/attachments/20091201/217e8fff/attachment-0001.png>
-------------- next part --------------




The results that I see are explained reasonably easily.

With TeX+dvi+GS the Postscript code produces small lines of colour
that are stroked for a particular length:  1.1 setlinewidth
and with equal linewidth:                  1.1 0 rlineto stroke
This gives the coloured squares that you see.


With pdfTeX, the coding is wrapped up differently:

12 0 obj <<
/Shading << /Sh << /ShadingType 1
/ColorSpace /DeviceRGB /Matrix [1 0 0 1 56.69363 56.69363]
/Domain [-56.69363 56.69363 -56.69363 56.69363] /Function 4 0 R >> >>

/ShadingType 1   sets the shading to be constant in the squares,
with the colour being calculated by  /Function 4 0 R
and the grid determined by the  /Domain  specification.

It would seem that Adobe handles the /ShadingType 1  a bit more
smoothly than does Preview. (Perhaps it has some extra
anti-aliasing at the edges of the squares?)


The conceptual difference between the PDF coding and the
PostScript coding is that with the PDF it is clear that we are
shading an extended region, so the result is supposed to look
smooth. (There is even the possibility of an extra setting
   /Antialias /True
which a viewer might implement, but doesn't have to.)

With the PostScript, on the other hand, it is really just
a collection of separate strokes in different colours.
The renderer does not have the extra hint that these are
meant to be filling a region. In particular there could
be some interaction with the pixel boundaries, leading
to a less-smooth result, visually.



The PDF Spec offers the following alternatives:


8.7.4.3   Shading Dictionaries

A shading dictionary specifies details of a particular gradient fill,  
including the type of shading to be used, the geometry of the area to  
be shaded, and the geometry of the gradient fill. Various shading  
types are available, depending on the value of the dictionary?s  
ShadingType entry:

?Function-based shadings (type 1) define the colour of every point in  
the domain using a mathematical function (not necessarily smooth or  
continuous).

?Axial shadings (type 2) define a colour blend along a line between  
two points, optionally extended beyond the boundary points by  
continuing the boundary colours.

?Radial shadings (type 3) define a blend between two circles,  
optionally extended beyond the boundary circles by continuing the  
boundary colours. This type of shading is commonly used to represent  
three-dimensional spheres and cones.

  ... and more ...


In an uncompressed version of the PDF you can simply edit
to replace  /ShadingType 1  by  /ShadingType 2 .
However Adobe Reader reports an error and shows no colour.

Preview, on the other hand, seems to be ignoring this parameter.
The colour remains, with no perceptible change.


But the /Function  needs to produce a different collection
of data, to suit the type 2 and type 3 shadings.
Tikz can use type 2 --- with  \pgfdeclarehorizontalshading
and \pgfdeclareverticalshading ---  and type 3 with
\pgfdeclareradialshading .
It would be interesting to see the results using these.

Perhaps the biggest problem with this, though, is that
whereas type 1 uses a single function of 2 parameters,
to assign colours within a 2-dimensional region,
types 2 and 3 use only a single parameter, varying in
a 1-dimensional way. So you would need to create a number
of radial strips (narrow sectors) and setup the shading
separately within each of these.
You would need to setup the sectors in TeX; or do a single
sector multiply, each shaded differently then rotated into
the correct place.

>
> But, for me, this is not very good : if you look closer, you can  
> see this (compiled with tex+dvi):
> <pgf4.png>
>
> Not smooth shade... I see big squarre (compare picture 1 and  
> picture 3 here : http://quark67.free.fr/pgf.html ).


Hope this helps.

And I'd still like to see what others get with later versions
of different viewer software.
How does it fit with the above analysis?


Cheers,

	Ross

------------------------------------------------------------------------
Ross Moore                                       ross at maths.mq.edu.au
Mathematics Department                           office: E7A-419
Macquarie University                             tel: +61 (0)2 9850 8955
Sydney, Australia  2109                          fax: +61 (0)2 9850 8114
------------------------------------------------------------------------