🚀 go-pugleaf

In article <1992Nov18.045514.4005@oucsace.cs.ohiou.edu> 
carlip@oucsace.cs.ohiou.edu (Walter C3arlip) writes:

>... As I understand it, a given style of type will
>have a different design depending upon the point size it is to be printed
>at, e.g., Times Roman at 12pt will have lines that are wider and darker
>relative to the hight of the letter than Times Roman at 20pt.

In the good old days of cold lead type (and maybe hot lead type, I don't know)
this was true.  A 'font' consisted of a certain typeface at a certain size,
so a different mold or type slug was used for each different size.
This in turn allowed the design to be varied among sizes in the way you 
describe.

When the industry switched to phototypesetters, this variation was lost.
The 'font' was then a collection of transparent masks at a single size, and
optics scaled the mask linearly to the desired size.

PostScript technology preserved this linear scaling.  The glyph outline
is described in bezier curves, and the PostScript interpreter scales the
outline linearly to the desired size.  Therefore, most PostScript typefaces
have a single design which is used for all sizes.

I say *most*, because Multiple Master font technology returns the 
flexibility type designers had with hot lead.  A Multiple Master typeface
contains two or more 'master designs', and you can create 'font instances'
which are a mixture of the two master designs.  If the designers choose,
they can make masters with two different 'design sizes'.  Adobe's Minion
Multiple Master typeface contains masters with optical sizes ranging from
(I think) 6 points to 72 points.

I don't claim to be Adobe's expert on type, but the above is a rough
explanation, and may show why you got two opposite answers to your question.

To give a practical answer to your practical question, I'd say that 
printing at a larger size with lower resolution and then photo-reducing 
down has always worked pretty well for me.  But then, I'm an engineer, not
a designer.

>_____________________________________________________________________________
>Walter C3arlip				    **** carlip@ace.cs.ohiou.edu ****
>(the "3" is silent)			

	--Jim DeLaHunt, Developer Support, Adobe Systems Incorporated	
delahunt@mv.us.adobe.com  ...!{apple,decwrl}!adobe!delahunt    +1-415-962-3790

In article <1992Nov18.181137.25836@adobe.com> delahunt@adobe.com (Jim DeLaHunt) writes:
>In article <1992Nov18.045514.4005@oucsace.cs.ohiou.edu> 
>carlip@oucsace.cs.ohiou.edu (Walter C3arlip) writes:

>>... As I understand it, a given style of type will
>>have a different design depending upon the point size it is to be printed
>>at, e.g., Times Roman at 12pt will have lines that are wider and darker
>>relative to the hight of the letter than Times Roman at 20pt.

>PostScript technology preserved this linear scaling.  The glyph outline
>is described in bezier curves, and the PostScript interpreter scales the
>outline linearly to the desired size.  Therefore, most PostScript typefaces
>have a single design which is used for all sizes.

I was under the impression that the "hints" allow the font machinery
to do adjustments based on target point size.  From my experiments, linear
font scaling doesn't look that great.

Psroff 3.0 has a HP Laserjet bitmap font scaling program.  It's strictly
linear.  When you scale up, the stroke relative widths stay constant.  But
what started being a reasonable Roman font starts looking bold, and bold starts
looking very fat.  (this is with relatively large magnifications.  Eg:
10pt -> 36pt)  Quite useable within my range of sizes, but still noticable.
-- 
Chris Lewis; clewis@ferret.ocunix.on.ca; Phone: Canada 613 832-0541
Psroff 3.0 info: psroff-request@ferret.ocunix.on.ca
Ferret list: ferret-request@ferret.ocunix.on.ca

In article <1992Nov18.045514.4005@oucsace.cs.ohiou.edu> carlip@oucsace.cs.ohiou.edu (Walter C3arlip) writes:

   I have a technical question to ask about Postscript Font Scaling and its
   relation to font design.  As I understand it, a given style of type will
   have a different design depending upon the point size it is to be printed
   at, e.g., Times Roman at 12pt will have lines that are wider and darker
   relative to the hight of the letter than Times Roman at 20pt.

   Now, when one uses postscript fonts, one has outline descriptions of the
   font.  Is it possible to determine what point size the font was designed
   for?  When one prints the font at a larger size, is the font scaled 
   linearly or is postscript smart enough to 'redesign' the font for the
   larger size? 

   Let me put this an another context.  Suppose one wants to produce 
   a journal using a 300dpi printer.  If one uses a 24pt design, and
   then uses photo reduction to 12pt, can one effectively get 600dpi
   resolution (I'd expect this to work if the fonts are scaled linearly),
   or will the reduced copy look like type designed for 24pt and
   then shrunk (so that the type does not match its design size)?

   I'm looking for the straight dope here, as I have already gotten 2 opposite
   answers to this from people I'd expect to know. 

Type 1 fonts are scaled linearly. So printing at 20pt and scaling 50%
essentially gives you the same shapes you would get if you had printed the
font at 10pt (modulo rasterization effects).

Most Type 1 fonts appear to have a design size of 10pt, 
although it is usually not specified.

Some Type 1 font families come with different fonts for different design
sizes.  For example the Computer Modern fonts in Type 1 form include cmr5,
cmr6, cmr7, cmr8, cmr9, cmr10, cmr12, and cmr17, where the numeric suffix
indicates the design point size.

So you certainly can have different designs for different point sizes, 
although you pay a price in terms of font downloading.

MultipleMaster fonts provide for continuous variation of some parameters,
although this is usually used for weight variation and condensed/expand
rather than design size.

Berthold K.P. Horn
Cambridge, Massachusetts, USA

In response to what Chris Lewis and Walter C3arlip wrote, and amplifying
what Jim Delahunt said, I want to add my $.02. As a caveat, let me say
I am not a type designer or expert, just an interested amateur.

    1.) The "design center" for scaling Type 1 fonts (except, of course,
        for *some* Multiple Master typefaces), is linear scaling at all
        point sizes.

    2.) Hints are used to overcome the limitations of low-resolution
        digital printers. They primarily prevent various kinds of uneven
        "color" and illegible characters. They do not have substantial
        effect at large point sizes and high resolutions.

    3.) The modification of a type design for use at different sizes,
        often called optical scaling, is one available axis of design
        in Multiple Master typefaces. This idea has been around a very
        long time, and is generally applied only to serif designs.
    4.) The reasons for different designs at different sizes in hot type
        (and Multiple Master Fonts) are complex and subtle:
        a.) With some printing techniques, especially older ones, ink
            does not stay where you want it, and type designs must be
            modified at smaller sizes to remain readable. This may mean
            simplification; it may also mean addition of strange
            features like inkwells to control ink placement.
        b.) For "readabilty" purposes in large blocks of text, it may
            be useful to sacrifice some of the uniqueness of a design.
            This may mean thickening serifs, lessening contrast between
            thin and thick strokes, increasing x-height, even changing
            the angles of bowls, all in the interest of smooth reading.
        c.) With physical pieces of metal, strength and integrity of
            the type through a long print run might dictate design
            changes, both at small and large sizes. This was especially
            true before the industrial revolution, as tooling hard
            metals was very difficult, and softer ones were used.

    5.) Some type designs have been intentionally tailored to the
        strengths and weaknesses of low-resolution printers. These
        designs might, indeed, appear "blocky" or "coarse" at higher
        resolutions or point sizes. Even "normal" fonts may not look as
        good at large sizes as faces specifically designed it. This is
        why, even before Multiple Masters, some fonts came with "Titling
        Caps". This is also why lighter weights are sometimes chosen for
        larger sizes (advertising use notwithstanding).

    6.) There is no reason "linear scaling" should look bad within a
        common-sense range of sizes. Hints will regularize things at
        small sizes, and the larger number of available pixels will make
        things look good at large sizes. The reasons for choosing a
        Multiple Master face should center on design flexibility,
        including better control of appearance at small sizes (for
        readability) and large sizes (for "elegance", faithfulness to
        the design, or whatever you want to label it.

123456789012345678901234567890123456789012345678901234567890123456789012
Stephen Zisk (zisk@adobe.com)  Disclaimer: These are my personal views
Adobe Systems Incorporated     and do not represent the position of
Eastern Regional Office        Adobe Systems Incorporated.
24 New England Exec. Park
Burlington, MA 01803           They pay me but they don't muzzle me!

clewis@ferret.ocunix.on.ca (Chris Lewis) writes:
>delahunt@adobe.com (Jim DeLaHunt) writes:
[metal type designs varied with size; phototypesetters linear]
>>PostScript technology preserved this linear scaling.  The glyph outline
>>is described in bezier curves, and the PostScript interpreter scales the
>>outline linearly to the desired size...
...
>I was under the impression that the "hints" allow the font machinery
>to do adjustments based on target point size.  From my experiments, linear
>font scaling doesn't look that great.

No, the hints are intended to help based on device resolution--they tell
the font-scaling code how to behave when the characters get small relative
to device resolution.  Examples of the hints:
	- "Flex" says "mash this curve flat if the "flex height" (roughly,
	  the deviation from a straight line) is too small relative to the
	  device pixel size.
	- "Stem width" hints say that when the size gets small enough that
	  a one-pixel difference in line width is noticeable, certain
	  widths should be made equal.
	- "Alignment zones" describe how to adjust horizontal segments so
	  those which should be at the same height actually are, when you
	  get down to where a one-pixel difference matters.
The point, again, is that they're all related to device resolution--they
kick in at 600 dpi at half the point size they do at 300.  So they don't
provide any help with the changes that should happen due to how our eyes
perceive the characters (i.e., changes based on actual size, irrespec-
tive of device resolution).

However, there are other effects that enter into apparent differences.  For
example, the rule for what pixels to turn on says "any pixel that a stroke
passes through.  This inherently darkens characters somewhat at small reso-
lutions.  (It's possible to compensate for this somewhat, but if the
correct width of a stroke is 0.5 pixel, your only choices are to make it
twice as wide as it should be or discard it and let the stroke break up.)
-- 
Dick Dunn    rcd@raven.eklektix.com   -or-   raven!rcd    Boulder, Colorado
	...Simpler is better.

In response to what Chris Lewis and Walter C3arlip wrote, and amplifying
what Jim Delahunt said, I want to add my $.02. As a caveat, let me say
I am not a type designer or expert, just an interested amateur.

    1.) The "design center" for scaling Type 1 fonts (except, of course,
        for *some* Multiple Master typefaces), is linear scaling at all
        point sizes.

    2.) Hints are used to overcome the limitations of low-resolution
        digital printers. They primarily prevent various kinds of uneven
        "color" and illegible characters. They do not have substantial
        effect at large point sizes and high resolutions.

    3.) The modification of a type design for use at different sizes,
        often called optical scaling, is one available axis of design
        in Multiple Master typefaces. This idea has been around a very
        long time, and is generally applied only to serif designs.
    4.) The reasons for different designs at different sizes in hot type
        (and Multiple Master Fonts) are complex and subtle:
        a.) With some printing techniques, especially older ones, ink
            does not stay where you want it, and type designs must be
            modified at smaller sizes to remain readable. This may mean
            simplification; it may also mean addition of strange
            features like inkwells to control ink placement.
        b.) For "readabilty" purposes in large blocks of text, it may
            be useful to sacrifice some of the uniqueness of a design.
            This may mean thickening serifs, lessening contrast between
            thin and thick strokes, increasing x-height, even changing
            the angles of bowls, all in the interest of smooth reading.
        c.) With physical pieces of metal, strength and integrity of
            the type through a long print run might dictate design
            changes, both at small and large sizes. This was especially
            true before the industrial revolution, as tooling hard
            metals was very difficult, and softer ones were used.

    5.) Some type designs have been intentionally tailored to the
        strengths and weaknesses of low-resolution printers. These
        designs might, indeed, appear "blocky" or "coarse" at higher
        resolutions or point sizes. Even "normal" fonts may not look as
        good at large sizes as faces specifically designed it. This is
        why, even before Multiple Masters, some fonts came with "Titling
        Caps". This is also why lighter weights are sometimes chosen for
        larger sizes (advertising use notwithstanding).

    6.) There is no reason "linear scaling" should look bad within a
        common-sense range of sizes. Hints will regularize things at
        small sizes, and the larger number of available pixels will make
        things look good at large sizes. The reasons for choosing a
        Multiple Master face should center on design flexibility,
        including better control of appearance at small sizes (for
        readability) and large sizes (for "elegance", faithfulness to
        the design, or whatever you want to label it.

Stephen Zisk (zisk@adobe.com)  Disclaimer: These are my personal views
Adobe Systems Incorporated     and do not represent the position of
Eastern Regional Office        Adobe Systems Incorporated.
24 New England Exec. Park
Burlington, MA 01803           They pay me but they don't muzzle me!

In article <4007@ecicrl.ocunix.on.ca> clewis@ferret.ocunix.on.ca (Chris Lewis) writes:
>I was under the impression that the "hints" allow the font machinery
>to do adjustments based on target point size.  From my experiments, linear
>font scaling doesn't look that great.

Adobe hints are used in rasterization of an outline. They try to
ensure that strokes have the same width, and so on. But they can't fix
an outline designed for 72 point to work well at 6 point. When you
reduce sizes, you need to make open spaces larger, thins relatively
thicker, and so on.

However, Bitstream's hints are much more intelligent, and in practice
alter the way they rasterize an outline at small point sizes to avoid
some of the problems with using an outline designed at a different size.
That's why Bitstream's fontware produces better looking output than
Adobe or Compugraphic.

--mark

In article <4007@ecicrl.ocunix.on.ca> clewis@ferret.ocunix.on.ca (Chris Lewis) writes:
>I was under the impression that the "hints" allow the font machinery
>to do adjustments based on target point size.  From my experiments, linear
>font scaling doesn't look that great.

The hints make adjustments based on the number of device pixel spanned by 
the scaled outline, not based on the point size per se.  Hinting will have
the same effect on a 10-point font on a 300 dpi device as on a 5-point font 
on a 600 dpi device.

>Chris Lewis; clewis@ferret.ocunix.on.ca; Phone: Canada 613 832-0541

	--Jim DeLaHunt, Developer Support, Adobe Systems Incorporated	
delahunt@mv.us.adobe.com  ...!{apple,decwrl}!adobe!delahunt    +1-415-962-3790