John Chambers wrote:
>Christophe writes:
>| I find PDF a good (if not perfect it's a lot better than GIF IMHO)
>| format for document exchange and I found useful to say that to help
>| users who want to exchange music with non abc litterate friends.
>
>Yeah; PDF (and PS) are a lot better than GIF or any other format that
>sends  the  scan lines.  PS and PDF draw lines and curves to the best
>resolution of the output device, so their quality is as good as  that
>device can produce.

Aint necessarily so.

When you export a GIF (or PNG, BMP, JPEG etc.) from a program which
displays music on the screen, what you get is exactly what is displayed
on screen.  If you send that to another user it will display on their
screen exactly as as it did on yours (maybe bigger or smaller if their
screen resolution is different, but with no distortion of the symbols).
If that user prints it out it will look the same as it does on screen:
OK, but not great, since it won't take advantage of the higher printer
resolution and won't look as crisp as you might expect.

When you export a Postscript file, or the PDF made from it, what you get
is the information which would be sent to a Postscript printer.  Send
that to another user and the results on screen are quite unpredictable,
depending on the software and settings used, the screen resolution
and colour depth etc.  The results will vary from almost unreadable
to OK, but not great.  The same applies if that user prints it on a
non-postscript printer.  If she prints it on a postscript printer the
results will always be excellent.

So "ps good, gif bad" is quite wrong.  It depends what they are going
to be used for.  Gif is safer, since it will always look the same as
it does to the sender, and will never be illegible.  PS (or PDF) can
give much higher quality under certain specific conditions, but can
also produce atrocious results when those conditions don't apply.

>GIF is only used because browsers understand it.

No, it's used because it's the most efficient way of compressing a
black and white (or 256 colour) picture into a small file.  The newer
PNG format is just as good, and free, but not yet as popular.  JPEG
is better for full-colour pictures and BMP is uncompressed, so yields
unnecessarily huge files.

>(Wouldn't it be useful if browsers would display PS and PDF?  As  far
>as  I can tell, the reason they don't is that PS and PDF are patented
>formats owned by Adobe. This ought not to matter, since it's legal to
>decode and display them. But it's easy to understand why people might
>be wary of doing something that has a high probability of getting  IP
>lawyers involved.  ;-)

It would be useful, but these would not replace gif or png in web
pages because the file sizes are much larger and the download times
that much longer, and because the results on-screen are unpredictable.


>| So I thought it could be useful to tell them to turn off antialiasing
>| (for example, with GSVIEW 4.* on a Windows box, go to the Media/Display
>| settings menu and set the Graphics Alpha to 1 bit).

Ho ho!  How is the naive user expected to understand the connection
between Graphics Alpha and antialiasing?

>With the version for unix/linus systems, it's the "State" menu, which
>has an "antialias" item.
>
>It's easy,  once  you  know  about  it.   But  I've  never  seen  any
>documentation on this, though I have dug around in the GV and GS docs
>quite a bit to learn about some other things.
>
>It's a bit odd that this would be on by default.  It also messes up a
>lot of text, though the damage isn't as bad as with music.  As far as
>I can tell, antialiasing is only useful with  images,  and  not  with
>very many of them.  So by default antialiasing should be off.

Antialiasing has it's uses.  Perhaps I should explain what it does, for
the benefit of the less-well informed.

When a computer is instructed to draw a line from position x1y1 to
position x2y2 on a display device, it must translate that information
into a row of coloured pixels.  If the line is horizontal or vertical,
the result looks like a clean line, as all the coloured pixels are
adjacent.  If the line is diagonal, the result will be a staircase,
which doesn't look so good.  If the instruction was not just for
a simple line, but for a pattern such as five equally-spaced horizontal
lines whose spacing is different from that of the screen pixels, some
distortion is inevitable, resulting in the lines being drawn different
distances apart.  The distortion is called aliasing because it's
exactly analagous to the aliasing which happens in audio when you
play two closely related frequencies together.

Graphics antialiasing is achieved by filling in the pixels which would
be partially included in the lines with shades of grey.  The staircase
effect on diagonal lines disappears because the steps have been filled
in with grey pixels, and the eye sees this as a smooth continuous line
(unless you use a magnifying glass to see how it's done).  With the
staff pattern however the results are less satisfactory, because lines
which happen to fall exactly on a row of pixels are unchanged, while
those that fall between two rows result in a double-width grey line.
(When antialiasing coloured pictures, it's somewhat more complicated,
since the colour of each pixel is a weighted average of all the
overlapping colours.)

The solution is to apply antialiasing selectively - draw orthogonal
lines so that they fall on pixel boundaries and don't need antialiasing,
and then apply antialising only to diagonal lines, curves and complex
symbols.  This is quite easy for the originating program to do, and
you can make a gif of the result which works anywhere.  It's not
possible for a program like Acrobat Reader do this though, so it
just antialiases everything.

Phil Taylor


To subscribe/unsubscribe, point your browser to: http://www.tullochgorm.com/lists.html

Reply via email to