Message-Id: <mailto:199507152031.PAA14790@library.wustl.edu> Date: Sat, 15 Jul 1995 16:36:07 -0400 From: S Merrell <mailto:jaxon@PANIX.COM> Subject: Re: Bit depth vs dpi To: Multiple recipients of list IMAGELIB
On Fri, 14 Jul 1995, Cynthia Frazier <mailto:cf12@CORNELL.EDU> said:>>When I teach the concepts of dots per inch (dpi) and bit depth,
>>I stress that each value affects the quality of the final output.
Yes. And the image of the cube is quite useful.. I use it with my students.
>>Most people think that printing to a laser writer requires the
>>equivalent in dpi to the resolution of the lase rprinter. In
>>fact, (and my experience here is limited to Apple printers)
>>images are printed in cells and therefore a 300 dpi printer
>>requires only about 150 dpi in the image.
You raise a third notion.. the printer's dpi (aka its addressible resolution). But when figuring how to print halftone images, you must also factor in the resolution of the requested output itself.. specifically, the screening frequency (lines per inch or LPI of printed dots) you ask the system to use when creating your halftone or color sep.
As you mention, halftone dots are created by cells of printer "spots" (the Apple LW has 300/inch). The number of spots in a cell equates to the maximum possible shades of gray each cell can represent. EG., if a cell is 4x4 (16 spots), the cell can create 17 different-sized halftone dots - 17 shades of gray. If the cells are 8x8, you can generate 65.
Thus, the number of grays you will obtain in your printed output depends on the total number of available spots (the printer's addressible resolution).. AND, on the number of cells used to generate dots. But the number of cells varies (printer dpi divided by the LPI being requested). IOW, the ratio of printer dpi to output LPI sets the number of grays that can be generated for a given printing situation. It's an inverse relationship... more LPI, fewer grays. Here's the formula: # of possible gray levels ˙printer dpi/screen LPI)^2 + 1.
This is why you can't get a quality halftone from a laserwriter... not enough addressible resolution to produce high screening frequencies at the same time as enough shades of gray to make the image appear continuous.
>>High-end printing can require information as high as 1200 to
>>2400 dpi and 32 bpp; however, if printing to a laser printer,
>>your needs may be much lower.
If the printer has an addressible resolution of 2400 dpi, solve the equation to derive maximum printable LPI for 256 shades of gray.. about 150 LPI.
The rule of thumb for generating halftones or color separations is 2 input pixels for every 1 LPI of screened output. Assuming your 2400 dpi printer (so that our 150 LPI haftone will have sufficient spots/cells to create 256 tones), halftoning at 150 LPI typically requires image data at ~300 dpi. There's a lot of debate about this ratio.. some say it can be as low as 1.2:1 - I've found lower ratios work ok only if the image is being printed large.
FWIW, your 300 dpi laserwriter driver defaults to ~53 LPI - which yields 33 shades of gray. Any more than 106 dpi image data means you're feeding the LW more data than it needs to produce the print (longer print times).
For more good discussion of all this, get Blatner & Roth's excellent _Real World Scanning and Halftones_, Peachpit Press (ISBN 1-56609-093-8). <http://www.peachpit.com>.
Non-American readers will please substitute dpm/lpm and divide by ~25.4...
With apologies for the length,
Sam Merrell ___________________________________________ Synthetic Imaging, Inc/NYC - 212.684.6311 mailto: mailto:jaxon@panix.com - 76702.1252@compuserve.com