Mike Frasier used Fracton to render Jim's FOTD for May 6th, 2013
Mike Frasier used his Fracton program (on a Mac) specifying ~28 decimal digits of precision and anti-aliasing to render a version of Jim's FOTD for May 6th, 2013 without loss of precision artifacts. To see the improvement, the two images to compare are Fractint's anti-aliased version, and Mike's version. You can also compare the unanti-aliased Fractint version with Mike's... Don't forget to click on the image in your browser to view it actual size Â- to prevent it from being re-sampled by your browser! If a browser re-samples an image, it can significantly change what you see and make comparisons between images meaningless. Please navigate to May 6, 2013 thru my home page: Thanks. http://www.emarketingiseasy.com/TESTS/FOTD/jim_muths_fotd.html Jim always uses: float=y and, assuming that Fractint uses IEEE 754 double-precision binary floating-point format for its floating point calculations, Fractint is using 15 to 17 decimal digits during its number crunching: http://en.wikipedia.org/wiki/Double-precision_floating-point_format // Would a Fractint developer please comment on my assumption? Of course, errors can accumulate during computation -- especially in repeated calculations as are used in fractals. I typically find visible anomalies starting to appear in Fractint fractals at magnifications of around 10^(11) or 10^(12) depending on the fractal. You can really see the improvement in Mike's ~28 decimal digit anti-aliased version clearly by using my "blink microscope" method (described at the bottom of each monthly page on my site of Jim's FOTDs.) The method is described where my page says: "Here's a nifty way to compare..." Please ignore the slight image shift when 'blinking' between the two versions of anti-aliased images. I'm really partial to using the keystroke method of switching the viewed tab -- since you don't have to take your eyes off the fractal to switch the tab you're viewing. The differences between the images will then hit you in the face like a wet mackerel. - Hal Lane ######################## # hallane@earthlink.net ########################
And how might the Fractint version look with more than 28 decimal digits of previous courtesy of using arbitrary precision? On 05/09/2013 12:52 PM, Hal Lane wrote:
Mike Frasier used his Fracton program (on a Mac) specifying ~28 decimal digits of precision and anti-aliasing to render a version of Jim's FOTD for May 6th, 2013 without loss of precision artifacts.
To see the improvement, the two images to compare are Fractint's anti-aliased version, and Mike's version. You can also compare the unanti-aliased Fractint version with Mike's...
Don't forget to click on the image in your browser to view it actual size –- to prevent it from being re-sampled by your browser! If a browser re-samples an image, it can significantly change what you see and make comparisons between images meaningless.
Please navigate to May 6, 2013 thru my home page: Thanks. http://www.emarketingiseasy.com/TESTS/FOTD/jim_muths_fotd.html
Jim always uses: float=y and, assuming that Fractint uses IEEE 754 double-precision binary floating-point format for its floating point calculations, Fractint is using 15 to 17 decimal digits during its number crunching: http://en.wikipedia.org/wiki/Double-precision_floating-point_format
// Would a Fractint developer please comment on my assumption?
Of course, errors can accumulate during computation -- especially in repeated calculations as are used in fractals. I typically find visible anomalies starting to appear in Fractint fractals at magnifications of around 10^(11) or 10^(12) depending on the fractal.
You can really see the improvement in Mike's ~28 decimal digit anti-aliased version clearly by using my "blink microscope" method (described at the bottom of each monthly page on my site of Jim's FOTDs.) The method is described where my page says: "Here's a nifty way to compare..."
Please ignore the slight image shift when 'blinking' between the two versions of anti-aliased images.
I'm really partial to using the keystroke method of switching the viewed tab -- since you don't have to take your eyes off the fractal to switch the tab you're viewing. The differences between the images will then hit you in the face like a wet mackerel.
- Hal Lane
######################## # hallane@earthlink.net ########################
_______________________________________________ Fractint mailing list Fractint@mailman.xmission.com http://mailman.xmission.com/cgi-bin/mailman/listinfo/fractint
-- David gnome@hawaii.rr.com authenticity, honesty, community http://dancingtreefrog.com http://clanjones.org/david/ http://dancing-treefrog.deviantart.com/
Sheesh ... more than 28 decimal digits of PRECISION couresy of ... Is it Frieday yet? On 05/09/2013 11:14 PM, david wrote:
And how might the Fractint version look with more than 28 decimal digits of previous courtesy of using arbitrary precision?
On 05/09/2013 12:52 PM, Hal Lane wrote:
Mike Frasier used his Fracton program (on a Mac) specifying ~28 decimal digits of precision and anti-aliasing to render a version of Jim's FOTD for May 6th, 2013 without loss of precision artifacts.
To see the improvement, the two images to compare are Fractint's anti-aliased version, and Mike's version. You can also compare the unanti-aliased Fractint version with Mike's...
Don't forget to click on the image in your browser to view it actual size –- to prevent it from being re-sampled by your browser! If a browser re-samples an image, it can significantly change what you see and make comparisons between images meaningless.
Please navigate to May 6, 2013 thru my home page: Thanks. http://www.emarketingiseasy.com/TESTS/FOTD/jim_muths_fotd.html
Jim always uses: float=y and, assuming that Fractint uses IEEE 754 double-precision binary floating-point format for its floating point calculations, Fractint is using 15 to 17 decimal digits during its number crunching: http://en.wikipedia.org/wiki/Double-precision_floating-point_format
// Would a Fractint developer please comment on my assumption?
Of course, errors can accumulate during computation -- especially in repeated calculations as are used in fractals. I typically find visible anomalies starting to appear in Fractint fractals at magnifications of around 10^(11) or 10^(12) depending on the fractal.
You can really see the improvement in Mike's ~28 decimal digit anti-aliased version clearly by using my "blink microscope" method (described at the bottom of each monthly page on my site of Jim's FOTDs.) The method is described where my page says: "Here's a nifty way to compare..."
Please ignore the slight image shift when 'blinking' between the two versions of anti-aliased images.
I'm really partial to using the keystroke method of switching the viewed tab -- since you don't have to take your eyes off the fractal to switch the tab you're viewing. The differences between the images will then hit you in the face like a wet mackerel.
- Hal Lane
######################## # hallane@earthlink.net ########################
-- David gnome@hawaii.rr.com authenticity, honesty, community http://dancingtreefrog.com http://clanjones.org/david/ http://dancing-treefrog.deviantart.com/
Unless I am mistaken, FractInt does arbitrary precision math only for Mandelbrot and Julia built in formula types. You can't select arbitrary precision math for formula types. -- Mike Frazier www.fracton.org
participants (3)
-
david -
Hal Lane -
Mike Frazier