|
David Makin
|
 |
« Reply #30 on: January 19, 2010, 12:09:08 AM » |
|
Thanks guys I got it working! May I suggest that the future defaults render a beautiful Mandelbulb?  I have a couple more questions. After trying all of the "Viewing control" parameters I was unable to find an appropriate control for zooming into the fractal. Is it true that using this software I can zoom inside the Mandelbulb? Also, I've read that UF supports arbitrary precision calculations, so i can continue zooming if I like? Perhaps I can make such exploration computationally tractable by first rendering at 50x50 and then increasing the resolution after I find something that looks interesting. you zoom in by double clicking ( or use options->mouse to configure ) and it is true, you can zoom very deep with ultrafractal, but make sure youve got enough time for waiting  i have asked frederik slijkerman, the programmer of uf, to include a virtual fractal size, to allow sizing of very small images, to have at least a big output Remember you can right-click on a parameter value control and use "Explore" to preview changes to the parameter using the small preview render that appears. |
|
|
|
|
Logged
|
|
|
|
|
David Makin
|
|
« Reply #31 on: January 19, 2010, 12:13:14 AM » |
|
ok, one more question,
and finally, how to set the iteration depth in general ? is it the standard iteration parameter ?!
No, for the 3D render use the "Max. Iterations" formula parameter. Use the standard UF Maximum Iterations parameter to control how much data is passed from the formula to the colouring. If using the usual colouring methods then a value of 7 is OK for this, if using "Lighting+UF colouring" then you should use 6+value where value is the number of iterations you wish to pass to the colouring - ideally value should be *more* than the number of iterations that have actually been used anywhere in the fractal render - typically 20 to 100. |
|
|
|
|
Logged
|
|
|
|
|
David Makin
|
|
« Reply #32 on: January 19, 2010, 12:18:53 AM » |
|
@dave what coloring method is used to color the SURFACE of the bulb ?
and what is the best approach to test some alternative coloring methods ?
i thing that some inside distance estimators would make good textures .... assuming that they have nice pattern in the inside section ....
The colouring method in the default is "Lighting+Iteration" - that uses smooth iteration for the surface colour - I think the other options for the "Colouring" parameter are fairly self-explanatory except I should say that when you select "Lighting+UF colouring" then you need to enable "UF Colouring" in the colouring formula and then you can plug any UF class cololring that you fancy *but* of course these are all currently based on complex (2D) numbers so there are parameters that appear in the 3D formula when you select "Lighting+UF colouring" that allow you to specify how the triplex (x,y,z) values are converted to complex (x,y) values for the colouring. |
|
|
|
|
Logged
|
|
|
|
|
David Makin
|
|
« Reply #33 on: January 19, 2010, 12:21:49 AM » |
|
how to rotate the bulb ?!!?
"Camera Rot." See the hints on the parameters  To rotate so the camera doesn't actually point directly at the current target then use "Camera Dir.". lol, didnt wanted to play with camera parameters, thx alot ! When there's only 1 object involved then camera rotation<=>object rotation Edit: Of course that's assuming you use the formula in the default manner where the target is the centre of the object. I have a lot more work to do on camera/object orientation/positioning  |
|
|
|
« Last Edit: January 19, 2010, 12:30:10 AM by David Makin »
|
Logged
|
|
|
|
|
Melancholyman
Guest
|
|
« Reply #34 on: January 19, 2010, 12:32:51 AM » |
|
Thx for all the help, I finally understood how to render in UF!! It would be nice with a 50ghz cpu though:(
|
|
|
|
|
Logged
|
|
|
|
|
cortexman
Guest
|
|
« Reply #35 on: January 19, 2010, 12:51:17 AM » |
|
Has anyone used the networking feature of UF? It seems like it should be possible to scale rendering speed of deep zooms linearly with number of machines. That sure would be nice..
Another question I have: Do you find mandelbulbs within mandelbulbs at arbitrary zooms as you do with the mandelbrot? So far I haven't been lucky enough to find one.
|
|
|
|
|
Logged
|
|
|
|
|
David Makin
|
|
« Reply #36 on: January 19, 2010, 01:10:21 AM » |
|
Has anyone used the networking feature of UF? It seems like it should be possible to scale rendering speed of deep zooms linearly with number of machines. That sure would be nice..
Another question I have: Do you find mandelbulbs within mandelbulbs at arbitrary zooms as you do with the mandelbrot? So far I haven't been lucky enough to find one.
I'm sure they are there *but* they seem to be well buried with small gaps around them and then more of the Mandelbulb - the best way to find them is to cut slices through the Mandelbulbs using the cutting planes. |
|
|
|
|
Logged
|
|
|
|
|
cKleinhuis
|
|
« Reply #37 on: January 19, 2010, 01:12:32 AM » |
|
i believe daves formula is not using global tables, and this would indeed speed up the distributed rendering ...
|
|
|
|
|
Logged
|
---
divide and conquer - iterate and rule - chaos is No random!
|
|
|
|
David Makin
|
|
« Reply #38 on: January 19, 2010, 01:22:16 AM » |
|
Thx for all the help, I finally understood how to render in UF!! It would be nice with a 50ghz cpu though:(
Hmmm - I get the default for my formula rendered using my 3GHz P4HT in around 35 to 50 secs @640*480 (yes my processor does vary that much - it's not cooled properly) - this would translate to around 2 to 3 secs on a dual quadcore system (that's why I'm saving for one). On a 2GHz core2duo then the default should be rendering @640*480 in around 10 to 17 secs Note that the only formulas that are optimised fully (until someone else finds a trick I haven't tried are the "White/Nylander" (+ or - sine) and the "Thornton 1". All the other "Bulbs" can be optimised for UF in a similar manner but it's quite a lot of work Edit: I should add that the really noticeable optimisation on the "White/Nylander" and "Thornton 1" is for integer powers from 2 to 31. |
|
|
|
« Last Edit: January 19, 2010, 01:24:58 AM by David Makin »
|
Logged
|
|
|
|
|
Melancholyman
Guest
|
|
« Reply #39 on: January 19, 2010, 01:55:59 AM » |
|
Thx for all the help, I finally understood how to render in UF!! It would be nice with a 50ghz cpu though:(
Hmmm - I get the default for my formula rendered using my 3GHz P4HT in around 35 to 50 secs @640*480 (yes my processor does vary that much - it's not cooled properly) - this would translate to around 2 to 3 secs on a dual quadcore system (that's why I'm saving for one). On a 2GHz core2duo then the default should be rendering @640*480 in around 10 to 17 secs Note that the only formulas that are optimised fully (until someone else finds a trick I haven't tried are the "White/Nylander" (+ or - sine) and the "Thornton 1". All the other "Bulbs" can be optimised for UF in a similar manner but it's quite a lot of work Edit: I should add that the really noticeable optimisation on the "White/Nylander" and "Thornton 1" is for integer powers from 2 to 31. The default formula on my core2duo 2,4GHz@3,2 renders in about 8 seconds, and with the formula you gave in this thread I render the "red" mandelbulb in about 55 seconds. But I want to zoom deeeep, and there it really starts to taking time  |
|
|
|
|
Logged
|
|
|
|
|
David Makin
|
|
« Reply #40 on: January 19, 2010, 02:17:04 AM » |
|
But I want to zoom deeeep, and there it really starts to taking time Ah - well there you're stuck with using the CPU at the moment in any case until someone writes GPU code that uses doubles |
|
|
|
|
Logged
|
|
|
|
|
Melancholyman
Guest
|
|
« Reply #41 on: January 19, 2010, 02:24:30 AM » |
|
But I want to zoom deeeep, and there it really starts to taking time Ah - well there you're stuck with using the CPU at the moment in any case until someone writes GPU code that uses doubles I would, if I could. I guess I have to take on the role as the "man in the field" exploring what new territories we discover, without fear of course.  The future is very interesting atm! |
|
|
|
|
Logged
|
|
|
|
|
cKleinhuis
|
|
« Reply #42 on: January 19, 2010, 02:30:13 AM » |
|
But I want to zoom deeeep, and there it really starts to taking time Ah - well there you're stuck with using the CPU at the moment in any case until someone writes GPU code that uses doubles doubles arent enough!!!!!!!!!!!!! |
|
|
|
|
Logged
|
---
divide and conquer - iterate and rule - chaos is No random!
|
|
|
|
Melancholyman
Guest
|
|
« Reply #43 on: January 19, 2010, 02:35:13 AM » |
|
But I want to zoom deeeep, and there it really starts to taking time Ah - well there you're stuck with using the CPU at the moment in any case until someone writes GPU code that uses doubles doubles arent enough!!!!!!!!!!!!! The new Nvidia graphic card (fermi) seems to have something in the realm of "raytracing". Maybe this will be a substrate (basis) for making incredible Mandelbulb implementations! |
|
|
|
|
Logged
|
|
|
|
|
David Makin
|
|
« Reply #44 on: January 19, 2010, 02:36:44 AM » |
|
But I want to zoom deeeep, and there it really starts to taking time Ah - well there you're stuck with using the CPU at the moment in any case until someone writes GPU code that uses doubles doubles arent enough!!!!!!!!!!!!! Well you can always use UF's arbitrary precision - if you're very, very patient - note that to do so you would need to set UF to use it "Always" and not just "When necessary". |
|
|
|
|
Logged
|
|
|
|
|
June 19, 2009, 03:01:21 AM
Dilber MC Theme by HarzeM