cnc@hpcid.HP.COM (Chris Christensen) (09/23/87)
I am looking for information about dithering algorithms. I often see images on the Mac that were dithered with an algorithm with which I am unfamiliar. I am familiar with contrained average (checkerboard like patterns) and pseudo-optical screening (various size dots), but the algorithm I speak of looks like it uses a pseudo random screen. It is grainy in apperence but has fewer "dithering artifacts" than either of the above algorithms. For anyone with Hypercard, the picture of the baby in the photo album stack uses the algorithm to which I refer. So... anyone know what I am talking about? -------------------------------------------------------------------------------- Chris Christensen Hewlett-Packard (408) 553-2955 or telnet 553-2955 Design Technology Center cnc@hpcid.hp.com Santa Clara, CA hplabs!hpda!cnc (disclaimer: my opinions are not even my own)
olson@endor.harvard.edu (Eric Olson) (09/25/87)
In article <3320020@hpcid.HP.COM> cnc@hpcid.HP.COM (Chris Christensen) writes: >I am looking for information about dithering algorithms. > >I often see images on the Mac that were dithered with an algorithm with >which I am unfamiliar. > >I am familiar with contrained average (checkerboard like patterns) and >pseudo-optical screening (various size dots), but the algorithm I speak >of looks like it uses a pseudo random screen. It is grainy in apperence >but has fewer "dithering artifacts" than either of the above algorithms. > >For anyone with Hypercard, the picture of the baby in the photo album stack >uses the algorithm to which I refer. That is the dither generated by the Koala MacVision slow-scan video digitizer. I believe its called "Floyd Steinberg Dither": this is what it is called in a very old version of GrayPaint (a/k/a The Realist a/k/a The Image Studio, now by Letraset). I don't know anything else about it. The MacVision was relatively inexpensive, and a very popular digitizer. I don't think it's available anymore. The baby in the Hypercard Slide Show stack is Bill Atkinson's daughter, Laura. -Eric Eric K. Olson olson@endor.harvard.edu harvard!endor!olson
tow@parcvax.Xerox.COM (Robert Francis Tow) (09/25/87)
In article <3320020@hpcid.HP.COM> cnc@hpcid.HP.COM (Chris Christensen) writes: > >I am looking for information about dithering algorithms. > >I often see images on the Mac that were dithered with an algorithm with >which I am unfamiliar. > >I am familiar with contrained average (checkerboard like patterns) and >pseudo-optical screening (various size dots), but the algorithm I speak >of looks like it uses a pseudo random screen. It is grainy in apperence >but has fewer "dithering artifacts" than either of the above algorithms. > >For anyone with Hypercard, the picture of the baby in the photo album stack >uses the algorithm to which I refer. > > >So... anyone know what I am talking about? > The Floyd-Steinberg error diffusion method is a better basic method, which results in images that do not exhibit posterization. In this class of algorithms, when a pixel is compared against a threshold value (which may be varied according to some screening or dithering pattern) and either set or not set in the output bitmap, the 'error' from this is distributed to neighboring pixels. Error diffusion algorithms tend to produce several artifacts peculiar to themselves: crystalline-like patterns, ringing at edges, wave like patterns, and a blurring of the image. There are proprietary versions of these algorithms which manage to avoid these (I have myself worked out some), but these tend to be closely guarded by scanner and graphics companies. You will find the basic Floyd-Steinberg error diffusion algorithm described on pages 104-106 of 'Procedural Elements of Computer Graphics' by David F. Rogers (MacGraw-Hill, 1985); the original paper is 'An Adaptive Algorithm for Spatial Gray Scale, SID 1975', Int. Symp. Dig. Tech., Pap., pp. 36-37, 1975, by Floyd and Steinberg. An excellent reference for halftoning and image processing techniques applicable to halftoning and dithering is 'Digital Document Processing', by Hsieh S. Hou (John Wiley and Sons, 1983). Hint: I found some quite useful starting points here which lead to a vastly improved version of the error diffusion algorithm. Rob Tow Graphics Wizard ("I can call spirits from the vasty deep") ParcPlace Systems 2400 Geng Road Palo Alto CA. 94303 ARPANet: tow@ParcPlace.com USENET: ...!{ucbvax, allegra, seismo, pyramid, decvax}!sun!pplace!tow