Article View: sci.fractals
Article #376Re: 24 BIT Fractal engine: Why not?
From: NOEL_GIFFIN
Date: Tue, 28 Feb 1995 23:15
Date: Tue, 28 Feb 1995 23:15
61 lines
5065 bytes
5065 bytes
In article <3j01u8$qqi@news.bu.edu>, genew@bu.edu (Doyle Williams) writes... >Greetings all, > I'm looking for a 24BIT fractal program that has the same basic features >as FRACTINT (i.e., color cycling, 3Ding, zooming, skewing, ability to plug in >formulas, etc.). This program could be freeware, shareware, or >commercial. I've got a couple of the generators that just do Mandelbrots, but >we're looking for more options (not that a Mandel is boring) and variety along >the lines of FRACTINT. >Thanks Again There are a few myth's and misconceptions that should be discussed in this thread. Some of them may be mine. First thing, 24 bit fractal generation itself hasn't really been properly implemented by anyone that I'm aware of. Ok, people tell me that Kai's Power Tools has a mandelbrot generator that can produce 24 bit images. From what I hear, and the images I have seen that it produced, it looks like it uses the iteration count to produce a 24 bit image by shading and filtering algorithms. I was told that colour control consists of 512 setpoints that will software interpolate to produce a full spectrum. Does anyone know if this is anywhere near the truth? Fine, so it does produce a 24 bit fractal image, but in this regard so does any raytracer that can import a height field and many nice 24 bit fractals have been produced this way, but that isn't really my idea of a 24 bit fractal generator. Someone just released a small low resolution 24 bit lyapunov program for ibmpc's but I haven't had time to look at it yet. Anyone care to review it? Maybe some of the Mac software might be 24 bit, I'm not a Mac person so I can't be sure, but, I don't think Mandella or Mandelzot or any of the well known Mac programs are 24 bit. There are some major technical problems with implementing 24 bit fractal generators. First of all basic colouring algorithms don't work as simply anymore. No one is going to iterate each pixel 4 million times, at least not with this current generation of computing hardware. Tim Wegner started a thread years ago in Alt.Fractals trying to discuss various colouring algorithms for 24 bit fractals. He had some nice ideas of combining colouring options such as iteration count, real component, imag , mult, summ, etc... Pick any three options and apply one to each of the Red, Green, and Blue portion to produce a 24 bit image. I think there is even some instructions in the current version of fractint to create this with existing software by calculating the image three times and then overlaying them with another software package. This is all do-able right now but perhaps a little cumbersome. Anyone got any nice images that they have created this way and want to show off? A real 24 bit fractal generator using these techniques would have to let you select 3 algorithms and then calculate each pixel three times. Well maybe only once if you chose only one escape criteria. There maybe a lot of different techniques out there that haven't been thought of. I think its worth a lot of discussion. Any more thoughts Tim? One thing that you can forget about is colour cycling in 24 bit. At least with the current generation of graphics cards. Colour cycling works in 8 bit because the pallette is loaded into a hardware table on the graphics card and the image data is mapped to RGB values by using the 8-bit pixel value as an index into the CLUT (colour lookup table). Colour cycling is achieved by shifting the pallette through the CLUT like a ring buffer. The image data doesn't change. Once you get into 16 bit and 24 bit images, things don't work this way anymore. Almost all 16 and 24 bit cards use the image data directly. Someone told me the 3DO cards have 16 CLUT. Is this true? If your using image data directly then colour cycling means scanning the image for all occurances of each pixel value present and replacing it with a new one. I guess you could do it in a single pass if you had a map of replacement values but it would have to be a large table and it would be done in software so you won't get the nice fast colour cycling that
Technical Details
Message-ID:
<28FEB199515153851@reg.triumf.ca>Path:
arc1-ca.pugleaf.net!nntp.gmd.de!dearn!blekul11!idefix.CS.kuleuven.ac.be!Belgium.EU.net!EU.net! howland.reston.ans.net!news.sprintlink.net!uunet!newsflash.concordia.ca!canopus.cc.umanitoba.ca!tribune.usask.ca!rover.ucs.ualberta.ca!unixg.ubc.ca!reg.triumf.ca!noelReferences:
<3j01u8$qqi@news.bu.edu>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: ff06e,dd9d3a316be96745,start X-Google-Attributes: gidff06e,public X-Google-Thread: 10ed18,a57e28850171a133 X-Google-Attributes: gid10ed18,public X-Google-ArrivalTime: 1995-02-28 19:57:24 PST Path: arc1-ca.pugleaf.net!nntp.gmd.de!dearn!blekul11!idefix.CS.kuleuven.ac.be!Belgium.EU.net!EU.net! howland.reston.ans.net!news.sprintlink.net!uunet!newsflash.concordia.ca!canopus.cc.umanitoba.ca!tribune.usask.ca!rover.ucs.ualberta.ca!unixg.ubc.ca!reg.triumf.ca!noel From: noel@reg.triumf.ca (NOEL_GIFFIN) Newsgroups: sci.fractals,bit.listserv.frac-l Subject: Re: 24 BIT Fractal engine: Why not? Date: 28 Feb 1995 15:15 PST Organization: TRIUMF: Tri-University Meson Facility Lines: 62 Distribution: world Message-ID: <28FEB199515153851@reg.triumf.ca> References: <3j01u8$qqi@news.bu.edu> NNTP-Posting-Host: reg.triumf.ca News-Software: VAX/VMS VNEWS 1.41