Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site ecsvax.UUCP Path: utzoo!watmath!clyde!bonnie!akgua!mcnc!ecsvax!bet From: bet@ecsvax.UUCP (Bennett E. Todd III) Newsgroups: net.graphics Subject: Re: Mandelbrot programs posted to net.sources Message-ID: <426@ecsvax.UUCP> Date: Mon, 16-Sep-85 17:18:29 EDT Article-I.D.: ecsvax.426 Posted: Mon Sep 16 17:18:29 1985 Date-Received: Wed, 18-Sep-85 05:42:00 EDT References: <417@ecsvax.UUCP> <2197@ukma.UUCP> Reply-To: duccpc!bet@ecsvax.UUCP (Bennett E. Todd III) Distribution: net Organization: Duke University Computation Center Lines: 56 In article <2197@ukma.UUCP> sean@ukma.UUCP (Sean Casey) writes: >Could you post some good sample values for input? Sorry 'bout that -- an obvious omission. Here are the headers from the files I have generated so far: Centered at -0.240000+0.000000i, scale 1.757800+1.528500i, 320x200 Centered at -0.720000+0.250000i, scale 0.019531+0.016984i, 320x200 Centered at -0.915000+0.270000i, scale 0.019531+0.016984i, 320x200 Centered at -1.410000+0.000000i, scale 0.019531+0.016984i, 320x200 Centered at -0.160000+1.030000i, scale 0.019531+0.016984i, 320x200 Centered at -0.550000+0.640000i, scale 0.019531+0.016984i, 320x200 Centered at -0.760000+0.110000i, scale 0.019531+0.016984i, 320x200 Centered at -0.240000+0.000000i, scale 1.757800+1.171866i, 1008x336 Centered at -0.760000+0.110000i, scale 0.025476+0.016984i, 1008x336 Centered at -0.720000+0.250000i, scale 0.025476+0.016984i, 1008x336 Centered at -0.915000+0.270000i, scale 0.025476+0.016984i, 1008x336 Centered at -1.410000+0.000000i, scale 0.025476+0.016984i, 1008x336 Centered at -0.160000+1.030000i, scale 0.025476+0.016984i, 1008x336 Centered at -0.550000+0.640000i, scale 0.025476+0.016984i, 1008x336 The 320x200 frames go nicely on the Color Graphics Adaptor; the 1008x336 come out about 4x6" on the Color Printer. The former take a few hours; the latter take more (one that I timed took 14 hours). This is with DeSmet C using the 8087 library; not optimal floating point performance by any means. Also, it looks to me like the zoomed in pictures on the Color Printer (90x zoom if I recall the numbers correctly) show pretty bad graininess around the border of the set. Either the double-precision 8087 support is slightly bad, or 255 isn't enough iterations. I fear the latter. Again, sorry about the omission. And to explain about the input parameters again, the command compute name center=a+bi scale=c+di hpix=e vpix=f creates a data file a little over hpix*vpix bytes long, for the picture centered at a+bi, going from a-c to a+c in the real direction and b-d to b+d in the imaginary, hpix pixels horizontally by vpix vertically. To cut down on typing I added a "like=name" option, so after computing one of the pictures you can get the scale, hpix, and vpix from it and just specify a new center by compute newname center=a+bi like=oldname As you can see, I should have put more effort into the algorithm and less into the user interface:-). Anybody have any concrete info on how many iterations are required to properly resolve the set at a given zoom? -Bennett -- "Hypocrisy is the vaseline of social intercourse." (Who said that?) Bennett Todd -- Duke Computation Center, Durham, NC 27706-7756; (919) 684-3695 UUCP: ...{decvax,seismo,philabs,ihnp4,akgua}!mcnc!ecsvax!duccpc!bet Brought to you by Super Global Mega Corp .com