Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!tut.cis.ohio-state.edu!pt.cs.cmu.edu!rochester!ken
From: ken@cs.rochester.edu (Ken Yap)
Newsgroups: comp.lang.postscript
Subject: Re: PostScript -- Stop Bashing Adobe
Message-ID: <1990Mar1.054811.15698@cs.rochester.edu>
Date: 1 Mar 90 05:48:11 GMT
References: <9466@imagen.UUCP>
Reply-To: ken@cs.rochester.edu
Organization: University of Rochester Computer Science Department
Lines: 39
Address: Rochester, NY 14627, (716) 275-1448

|I was hoping that some engineer or scientist from an institution such as
|AT&T Bell Laboratories, who is familiar with both PostScript and Claude
|Shannon's theory of information, would comment on the efficiency of
|PostScript.  What I would like to know is whether Bell Labs would accept
|PostScript as designed by Adobe Systems as the ONLY world standard for 
|the storing and transmitting of image data, and whether people at Bell
|Labs think that we need a more compact representation of PostScript 
|statements.

Why does it have to be somebody from Bell Labs? Any communications
engineer can understand the principles.

As far as the "inefficiency" of PS goes, there are two orthogonal
issues here: (1) How much does the 7 bit channel slow you down and (2)
how much can you squeeze the image data down to.

1. Encoding in hex will double the size of images compared to pure
binary.  If you get smart you could squeeze 4 bytes into 5 with a base
85 encoding, giving an efficiency of 80%. If it weren't for the control
characters, the theoretical maximum of the 7 bit channel would be 7/8th
of that of the 8 bit channel (87%). Not exactly a tremendous gain over
base 85 encoding.

2. How much information is really in the picture depends what's in the
picture.  Compression techniques such as run-length encoding, Huffman
encoding, etc, etc.  increase the information content by squeezing out
redundancy. A more aggresive technique is to retain the high-level
description of the picture, e.g. describe a circle by its radius and
position instead of sending the bitmap. Surprise, this is what some PS
programs do. Personally I'm looking forward to see if claims for
fractal encoding (recent SciAm) can be upheld.

So you trade-off processing time against transmission time.

By the way, PS is hardly a candidate as the ONLY world standard for
storing images. Ask people in graphics. I see it as a PDL, that's all.

If you want to flame this technical writeup, please do. It would be a
welcome change from the recent round of postings.