Subject:
|
Re: PostScript/EPS (was Re: Scans of My Own Train Stickers Wanted)
|
Newsgroups:
|
lugnet.off-topic.geek
|
Date:
|
Tue, 5 Jun 2001 04:51:26 GMT
|
Viewed:
|
139 times
|
| |
|
|
Wow! Thanks for that info Matt. (and thanks to Dan Boger who sent me a note
offline) PS/EPS are nifty.
I can see the value in using PS or EPS when *you* are the image source.
But I still have a question about its use as the payload carrier for scans.
If I make a scan of something, unless I subject it to some heavy processing
to determine edges and regions (which is prone to error) and some OCR to
read all the characters (which is prone to error), and some pattern matching
to determine what fonts they are (which is prone to error), isn't it still a
raster? That is, it is a collection of positions and color values, not a
vectorised collection of perfect geometry and region fills and font
callouts. (1) (2)
Right?
That is, if it gets rendered on a device that has a different dpi than the
scanner, either you have to print it at a different size, or you have to
interpolate. There was a long discussion either here, or in .publish, about
the proper resolution to do scans at. It was a long time ago and I may be
misremembering but I thought the general gist was to use nice even numbers,
like 300 dpi, rather than less readily scalable ones like 72 dpi, because
when you, for example, scale 72 to 600, the interpolation is worse than if
you scale 75 or 150 or 300 to 600. And I suspect transforming 72 to 3500 is
worse than transforming 75 to 3500. But I dunno.
Summary: Agree that EPS has huge advantages if you're the source of the
image. Still would love to hear more pro/con if the source is a scanner.
<snip>
++Lar
1 - as an aside, something similar to the raster/vector problem plagues
LDraw. LDraw parts are recorded using collections of points that denote
endpoints of segments that form the vertices of faces, rather than being
recorded as constructed solids which are made from perfect geometrical
forms. Thus they render poorly at extreme magnification (due to too few
approximating lines) and slowly at small scale (due to too many irrelevant
lines).
Were LDraw to be architected afresh, no doubt a geometric solid approach
would be taken, but it is way out of the question to do that at this point.
2 - as an another aside... consider flash. Flash movies that are bitmapped
images download/play much more slowly than flash movies that are geometric
descriptions. bitmaps are easier to make, they just take a scanner, but
geometric construction pays off in transmission time.
|
|
Message is in Reply To:
3 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
