wayner@kama.cs.cornell.edu (Peter Wayner) (11/09/90)
Here's an idea that someone has probably thought of already. When you want to use offset printing, the copy is photographically reproduced on the plate. This can involve some amount of enlarging and reduction. To generate 600 dots per inch, you can simply take a 300 dots per inch laser copy output and shrink it the right amount. (Roughly speaking). One problem with this is that the fonts are tuned to the action of the eye so 24 point letter is just not a 12 point letter expanded by a factor of two. The widths and the stroke lengths are tweaked by the font designer to provided two different collections of letters that look harmonious in the different sizes. The differences are quite large. My question is, would it be possible to print out a document that has been electronically enlarged. That is, twelve point font descriptions are printed twice as big, so when they are photographically reduced the resolution is (roughly) doubled? Is there anyone who has tried this? Is there any reason not to do it? -Peter Wayner (wayner@cs.cornell.edu) Peter Wayner Department of Computer Science Cornell Univ. Ithaca, NY 14850 EMail:wayner@cs.cornell.edu Office: 607-255-9202 or 255-1008 Home: 116 Oak Ave, Ithaca, NY 14850 Phone: 607-277-6678
p_davis@epik.enet.dec.com (Peter Davis) (11/09/90)
In article <48088@cornell.UUCP>, wayner@kama.cs.cornell.edu (Peter Wayner) writes... > >My question is, would it be possible to print out a document that has been >electronically enlarged. That is, twelve point font descriptions are printed >twice as big, so when they are photographically reduced the resolution >is (roughly) doubled? > Well, PostScript uses a single "master" (ie, character shape definition) for all point sizes, so you don't actually get different shapes for different sizes. Therefore, you can easily take a 300dpi PostScript printout, stat it down, and treat it as 600 dpi. Other than this, I don't know of any systems which do deal with multiple masters, except for phototypesetters. If you have a phototypesetter, you're already getting high resolution output.
wayner@kama.cs.cornell.edu (Peter Wayner) (11/09/90)
p_davis@epik.enet.dec.com (Peter Davis) writes: >In article <48088@cornell.UUCP>, wayner@kama.cs.cornell.edu (Peter Wayner) writes... >> >>My question is, would it be possible to print out a document that has been >>electronically enlarged. That is, twelve point font descriptions are printed >>twice as big, so when they are photographically reduced the resolution >>is (roughly) doubled? >> >Well, PostScript uses a single "master" (ie, character shape definition) for all >point sizes, so you don't actually get different shapes for different sizes. >Therefore, you can easily take a 300dpi PostScript printout, stat it down, and >treat it as 600 dpi. Does it really? Adobe must produce fonts that are tuned to different sizes. Can they be manipulated? -Peter Peter Wayner Department of Computer Science Cornell Univ. Ithaca, NY 14850 EMail:wayner@cs.cornell.edu Office: 607-255-9202 or 255-1008 Home: 116 Oak Ave, Ithaca, NY 14850 Phone: 607-277-6678
dhoyt@vz.acs.umn.edu (11/09/90)
.. Most postscript fonts do not have multiple rederings for different sizes. Unfortunate, but it does make it cheaper. So in theory, photoreduction can be used to increase the density of dots. However, to get a high quality image, you'll want a stat camera. Which would be expensive. A xerox or 35mm camera wouldn't do the job. It would probably be just as cheap, especially if you add in the cost of your time, to print the pages out on a lino or agfa. david
geof@aurora.com (Geoffrey H. Cooper) (11/10/90)
In article <48104@cornell.UUCP> wayner@kama.cs.cornell.edu (Peter Wayner) writes: >p_davis@epik.enet.dec.com (Peter Davis) writes: > > >>In article <48088@cornell.UUCP>, wayner@kama.cs.cornell.edu (Peter Wayner) writes... >>Well, PostScript uses a single "master" (ie, character shape definition) for all > >Does it really? Adobe must produce fonts that are tuned to different >sizes. Can they be manipulated? Actually, many phototypesetters also use a lense to photographically enlarge type. The day of design size versus actual size has passed to a large extent. However, Adobe does sell alternate versions of some fonts that are based on different masters. For example, you can get Times and Helvetica based on what look like 10 point masters. These characters have shapes more appropriate for 10 point, and look somewhat stodgy at larger sizes. There are also some type styles designed for classified adds that are intended for very small print size. But graphic arts is responding to the new concepts. The other day I saw a magazine article that was set in one of these classified add typefaces, at 12 point, to give the page a wide and spacy look. The article was about "the conflation of luxe & loose" clothes (I looked it up because of a reference in a Dave Barry article while killing time... really!), so I guess the typesetter decided to flaunt loose text with loose clothes and the loose-looking men that were in them. Had it not been for the humor of Dave Barry that put it all in proper perspective, I might have loosed my breakfast. - Geof -- geof@aurora.com / aurora!geof@decwrl.dec.com / geof%aurora.com@decwrl.dec.com
rberlin@birdland.sun.com (Rich Berlin) (11/10/90)
> Most postscript fonts do not have multiple rederings for different sizes. > Unfortunate, but it does make it cheaper. I've seen this several times. Are you all *sure* about this? My understanding is that Adobe Type 1 fonts (e.g. the built-ins on the LaserWriter) are "hinted," which means for example that at small sizes some of the strokes will be narrower than at large sizes. (To do it right, I think the metrics have to vary somewhat as well, but I don't know if the hinting does that.) -- Rich
prc@erbe.se (Robert Claeson) (11/11/90)
In a recent article dhoyt@vz.acs.umn.edu writes: > Most postscript fonts do not have multiple rederings for different sizes. >Unfortunate, but it does make it cheaper. So in theory, photoreduction >can be used to increase the density of dots. However, to get a high quality >image, you'll want a stat camera. Which would be expensive. A xerox or 35mm >camera wouldn't do the job. I beg to differ. I have successfully shrunk pages using a 35mm camera of good quality (a Canon EOS-1 with a Canon 50mm macro lens). Just make sure that the light is even. Use Kodak Technical Pan and develop in Tetenal Doku or Kodak Technidol. Gives a grainless result with normal contrast. The lens, not the film, is the limit using this setup. Followups to rec.photo, please. -- Robert Claeson |Reasonable mailers: rclaeson@erbe.se ERBE DATA AB | Dumb mailers: rclaeson%erbe.se@sunet.se Jakobsberg, Sweden | Perverse mailers: rclaeson%erbe.se@encore.com Any opinions expressed herein definitely belongs to me and not to my employer.
jfr@locus.com (Jon Rosen) (11/13/90)
In article <48104@cornell.UUCP> wayner@kama.cs.cornell.edu (Peter Wayner) writes: >>Well, PostScript uses a single "master" (ie, character shape definition) for all >>point sizes, so you don't actually get different shapes for different sizes. >>Therefore, you can easily take a 300dpi PostScript printout, stat it down, and >>treat it as 600 dpi. > >Does it really? Adobe must produce fonts that are tuned to different >sizes. Can they be manipulated? > Adobe's font sizing method with Postscript is very interesting. It is true that the base for all Type 1 fonts is a single set of stroke definitions that are enlarged and rasterized for any request point size. This allows theoretically unlimited sizing of all fonts. Also, the rasterizing process is done at the actual Postscript machine's resolution so that the rendering is optimal for each different type of printer. However, a strict mathematical rendering of the font definition in all sizes will not, as anyone familiar with typography would know, give a good "eyepleasing" typeface. Type 1 fonts in Adobe Postscript include "hints" which are essentially little pieces of Postscript code that are attached to each character definition and which modify the actual raster image depending on the size of the character, the style and possibly other variables like kerning, leading, etc. This means that as you enlarge fonts, you will see small changes in the actual raster image, mostly in the thickness of the body and stem elements and in the ascender/descender elements. This allows Adobe Type 1 fonts to look much better than simple magnification. It also means that sometimes simple overmagnification followed by stat reductions will not provide as much benefit as would be expected, and in some cases, may actually be counterproductive. Unfortunately, as far as I can tell, it is not simple to get at the hint definitions for modification. Jon Rosen
khcg0492@uxa.cso.uiuc.edu (Kenneth Holden Chang) (11/14/90)
In article <19458@oolong.la.locus.com> jfr@locus.com (Jon Rosen) writes: >Type 1 >fonts in Adobe Postscript include "hints" which are essentially little >pieces of Postscript code that are attached to each character definition >and which modify the actual raster image depending on the size of the >character, the style and possibly other variables like kerning, leading, >etc. This means that as you enlarge fonts, you will see small changes >in the actual raster image, mostly in the thickness of the body and stem >elements and in the ascender/descender elements. Correct me if I'm wrong, but this is *not* what the hints do, and is indeed one of the drawbacks of Postscript. What the "hints" do is adjust for low resolution (300 dpi) printers. Otherwise the letters tend to look "heavy." On a high-resolution printer like a Linotronic, however, fonts are scaled exactly, that is a 48 point 'A' looks exactly like a 24 point 'A' except it's larger. This is also Adobe's Expert Collections contain what are called "Titling" characters, i.e. fonts designed for larger point sizes. One of the advantages of TrueType when it finally appears is that it will incorporate the "nonlinear scaling" that Mr. Rosen describes. -- ***************************************************************************** Kenneth Chang * khcg0492@uxa.cso.uiuc.edu Center for Complex Systems Research * or University of Illinois * kc@complex.ccsr.uiuc.edu *****************************************************************************
bobs@sco.COM (Bob Stayton) (11/14/90)
We have done a lot of "over size" printing on standard 300
dpi printers and then reducing it to 75% with the goal of
improving the resolution. It certainly works, and the
results are quite a bit better than printing at the small
size (we are talking 10 pt finished size here). But the
letters do *not* look the same as 300 dpi or even 600 dpi
letters printed directly in that size. They look lighter,
thinner. We noticed it once when we mixed stuff printed
large with text printed regular. It was bad enough to
distract our readers when reproduced. As long as you are
consistent in how you print, though, it should work.
It is weird always working in oversize dimensions, though.
Also, unless your finished page size is very small, you
won't get 600 dpi equivalent, which requires a reduction to
50%. If you start with 8.5" x 11", a 50% reduction would
give 4.25" x 5.5", which is quite small (reduction goes in
both dimensions). If, like us, you are printing on half
pages (5.5" x 8.5"), you will get about a 75% reduction
(divide the long dimension after reduction by the long
dimension before reduction: 8.5 / 11 = 77%) .
bobs
Bob Stayton 425 Encinal Street
Technical Publications Santa Cruz, CA 95060
The Santa Cruz Operation, Inc. (408) 425-7222
...!uunet!sco!bobs
/* I don't speak for my company and they don't speak for me. */rcd@ico.isc.com (Dick Dunn) (11/14/90)
rberlin@birdland.sun.com (Rich Berlin) writes: > > Most postscript fonts do not have multiple rederings for different sizes. > > Unfortunate, but it does make it cheaper. > I've seen this several times. Are you all *sure* about this? My > understanding is that Adobe Type 1 fonts (e.g. the built-ins on the > LaserWriter) are "hinted," which means for example that at small sizes > some of the strokes will be narrower than at large sizes... There is adjustment of stroke widths and positions to make them look right on devices with "coarse" resolution. But that's not what the other folks were thinking of. What they had in mind was actual variations in stroke weights depending on the point size (independent of the device) to make letters "look right" at different sizes. Incidentally, one of the reasons for not doing this in PostScript fonts (aside from the obvious big one of cost!) is that there's a sort of nebulous relationship between the point size used to scale the font and what it means in final output...consider 0.5 dup scale /Times-Roman findfont 16 scalefont setfont If there were multiple designs depending on size, should this use the 16-pt artwork (the requested size for scalefont) or the 8-pt artwork (the actual output size)? It's even worse if you're going to do any enlargement or reduction after output, because it's the final size, as seen by the human eye, that determines which artwork to use. -- Dick Dunn rcd@ico.isc.com -or- ico!rcd Boulder, CO (303)449-2870 Cellular phones: more deadly than marijuana.
cs325ax@ux1.cso.uiuc.edu (David Hull) (11/14/90)
p_davis@epik.enet.dec.com (Peter Davis) writes: >Other than this, I don't know of any systems which do deal with multiple >masters, except for phototypesetters. If you have a phototypesetter, you're >already getting high resolution output. A properly designed Metafont font (as used by TeX) will have different feature sizes depending on its size. According to Don Knuth's book on the Computer Modern typefaces the vertical dimensions of Computer Modern Roman as scaled linearly, but "horizontally the smaller fonts are more extended, and they have relatively heavier weight (especially in hairlines)." -David
korp@atlantis.ees.anl.gov (Peter Korp) (11/14/90)
In article <1990Nov13.220222.14142@ico.isc.com> rcd@ico.isc.com (Dick Dunn) writes: [stuff deleted] >Incidentally, one of the reasons for not doing this in PostScript fonts >(aside from the obvious big one of cost!) is that there's a sort of >nebulous relationship between the point size used to scale the font and >what it means in final output...consider > 0.5 dup scale > /Times-Roman findfont 16 scalefont setfont >If there were multiple designs depending on size, should this use the 16-pt >artwork (the requested size for scalefont) or the 8-pt artwork (the actual >output size)? Pardon my stupidity, but now that you bring that point up which artwork does PostScript really use? I've used this stuff for years and never thought about this. >-- >Dick Dunn rcd@ico.isc.com -or- ico!rcd Boulder, CO (303)449-2870 > Cellular phones: more deadly than marijuana. Peter
pgd@bbt.se (11/17/90)
In article <11882@scorn.sco.COM> bobs (Bob Stayton) writes: > >We have done a lot of "over size" printing on standard 300 >dpi printers and then reducing it to 75% with the goal of >improving the resolution. It certainly works, and the >results are quite a bit better than printing at the small >size (we are talking 10 pt finished size here). But the >letters do *not* look the same as 300 dpi or even 600 dpi >letters printed directly in that size. They look lighter, >thinner. Our experience is that the photoreduction actually gives better result, than a similar higher-resolution laser-printer. It also gives better result than a similar higher-resolution imagesetter. This last is probably due to the blurring propensity of the laserprinter. One propensity of the laser-printers is that the letters are significantly emboldened. If you comparing the output from a (let's say) Linotronic, with a laserprinter, you will see the difference. So your thinning effect is probably just showing how the character are designed to look. Unless you have some problem in your photoreduction. (Like over-exposing)
pgd@bbt.se (11/18/90)
In article <19458@oolong.la.locus.com> jfr@locus.com (Jon Rosen) writes: >Unfortunately, as far as I can tell, it is not simple to get at the >hint definitions for modification. > You can get at the information, but who understands it, to modify it? Without a font-editor, that understands hints, it is very hard to do anything about it. (Does such an animal exist outside of Adobe labs?)
shiva@well.sf.ca.us (Kenneth Porter) (11/21/90)
I've always wondered what a Type 1 hinted font uses to decide what "real" size it should attempt to optimize for. Does it use the font's scale factor, the graphics state scale factor, or the product of these? If it only uses the font's scale factor, then one can scale the coordinate system and do photographic reduction to increase resolution. This fails if an application is using a graphic state scale factor to enforce its own coordinate system. Is there a device parameter that the font uses to decide the physical size of a character, such as a dpi variable? If one could change this, then photographic resolution enhancement becomes (computationally, not necessarily optically) trivial. Ken (shiva@well.sf.ca.us)
rcd@ico.isc.com (Dick Dunn) (11/27/90)
shiva@well.sf.ca.us (Kenneth Porter) writes: > I've always wondered what a Type 1 hinted font uses to decide > what "real" size it should attempt to optimize for. Does it use > the font's scale factor, the graphics state scale factor, or > the product of these?... Existing Type 1 fonts don't have any such "real size" optimizations. The tuning which is done is performed by the raster-conversion algorithms in the interpreter--but this is tuning for a particular device (and its par- ticular resolution and rendering characteristics), not for effects related to human perception and/or esthetics. -- Dick Dunn rcd@ico.isc.com -or- ico!rcd Boulder, CO (303)449-2870 ...Mr. Natural says, "Use the right tool for the job."