Re: Fractint Digest, Vol 25, Issue 10
This is where I think a shared internet or intranet cpu system could work. I was thinking a redesign of Fractint might be the solution but I'm moving completely out of the box. Consider this... When we do a fractal screen we see a colored image. But the underlying fractal is a coordinate plane of calculated values. Why not then be able to calculate those values and save them in a file. How different this is than disk video I don't know, but I do know I want a different effect. After the coordinate plane values is calculated it is retained. Then in fractint we "generate" the image again, but this time instead of calculating each point, they are simply read from the file. In this way we "play back" the image. We could even do the special effects while drawing the file and could redraw it at any speed we wanted. We could color or recolor the image any way we want. Essentially we could do anything that doesn't require a recalculation of any point. Since the file is just numbers calculated based on their location in the coordinate plane, that lends itself to divide and conquer. We decide how fine to dice the image up, and all those chunks are sent out to the grid and eventually returned filled in. There could even be a way to display the squares as they came back just for kicks. Now not knowing a lot about the internals for Fractint, does something like this seem feasible? It seems to me that the app to distribute the fractal around the net is then a separate application, and that fractint would only have to be modified to be able to read the file instead of calculating the points itself. -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.8.4 - Release Date: 3/27/2005 -- No virus found in this outgoing message. Checked by AVG Anti-Virus. Version: 7.0.308 / Virus Database: 266.8.4 - Release Date: 3/27/2005
Vortex Swirling wrote:
This is where I think a shared internet or intranet cpu system could work.
I was thinking a redesign of Fractint might be the solution but I'm moving completely out of the box. Consider this...
When we do a fractal screen we see a colored image. But the underlying fractal is a coordinate plane of calculated values. Why not then be able to calculate those values and save them in a file. How different this is than disk video I don't know, but I do know I want a different effect.
After the coordinate plane values is calculated it is retained. Then in fractint we "generate" the image again, but this time instead of calculating each point, they are simply read from the file. In this way we "play back" the image. We could even do the special effects while drawing the file and could redraw it at any speed we wanted. We could color or recolor the image any way we want. Essentially we could do anything that doesn't require a recalculation of any point.
Since the file is just numbers calculated based on their location in the coordinate plane, that lends itself to divide and conquer. We decide how fine to dice the image up, and all those chunks are sent out to the grid and eventually returned filled in. There could even be a way to display the squares as they came back just for kicks.
Now not knowing a lot about the internals for Fractint, does something like this seem feasible? It seems to me that the app to distribute the fractal around the net is then a separate application, and that fractint would only have to be modified to be able to read the file instead of calculating the points itself.
I can point out a couple of flaws with this concept to start with. The "coordinate plane of calculated values", that you refer to is in reality, a mesh or grid. An arbitrary net thown over an infinite sea of possible numbers, most of which are ill contained within the fixed precision of standard computing floating point numbers. Any saved set of calculations is very unlikely to be referenced again unless the exact starting co-ordinates, the magnification, iteration limit, and numerical precision were used. A very unlikely scenario. a small zoom factor/rotation/co-ordinate shift and you will not likely use more than a very few of your previously calculated values. The disk storage space required to store all possible floating point values of 80 bit numbers or even 64 bit numbers is prohibitive. 2 to the 64th = 1.84467x10e19 Multiply that by 8 bytes/value and that gives you a requirement for about 147 million terrabytes if my rough calculations are accurate. This speaks nothing of any arbitrary precision math or even changing the maximum iteration setting for the calculation, or heaven forbid, that you actually wanted to calculate a different fractal formula. regards, Noel Giffin
On Tue, 2005-03-29 at 17:16, Noel Giffin wrote:
Vortex Swirling wrote:
This is where I think a shared internet or intranet cpu system could work.
... I can point out a couple of flaws with this concept to start with.
Noel's right, calculating all the points in the set wouldn't work too well. There are several ways a network aware fractal generating program could provide cool additional functionality, though. 1) Speeding up calculations. Each user submits requests to the system, which are processed by the spare CPU cycles of other users whose computers are currently idle. This approach is used by Seti@home and the ElectricSheep screensaver, which produces flame-based animations. 2) Online repository of formulas, as in UltraFractal's formula database. The software can automatically download missing formulas. This could be expanded to .par and .map files as well. 3) A shared map of the M-set. Imagine if you could see the 'trails' of where other users had explored the set as rectangles surrounding the areas on the screen they'd been to, with names if they wanted to christen a particular valley or minibrot. (A central server would record the size and coordinates of each section each user renders.) You could follow their footsteps to cool areas or branch off to areas no-one had ever been before. -- Edwin
participants (3)
-
Edwin -
Noel Giffin -
Vortex Swirling