HOME | DD

Description
Banding-free spectral image based on goniometric rotation. There is some 'highlighting' of primary green and secondary orange, but that is to be expected. In comparison with some of te cruder algorithms it is still greatly suppressed, which is exactly what this 'weighed' color shifting is about.

There is an issue with rgbscript however: with most field numbers (the field number is set by the user according to the expected number of colors to be generated) it leaves stripes in the image. Finding a number of fields with which it doesn't is a bit of a lucky strike. In this case the color shift was by 1 degree, which should fill 360 fields exactly. The stripes only disappeared with 363 fields, meaning there is some overlap of top and bottom color (the script keeps pumping out spectra as long as there are fields to be filled).

Even weirder: this picture was created by setting the start color to green, but with output colors inverted. When I set the start color to red, with no inversion, but with the same 363 field count, the stripes are there again. Which suggests that they might be produced by something other than field height rounding. And of course, as usual, the resized JPEG shown on this page will probably display a ton of stripes, but of a different origin. A PNG should never show them, unless they are a part of the image itself, or the image viewer is indeed of a very crappy quality. By my knowledge that is..

Also: the lack of a proper yellow does annoy me. The algorithm plays safe, and in doing so, ends up having no personality whatsoever. How human.

Description
Banding-free spectral image based on goniometric rotation. There is some 'highlighting' of primary green and secondary orange, but that is to be expected. In comparison with some of te cruder algorithms it is still greatly suppressed, which is exactly what this 'weighed' color shifting is about.

There is an issue with rgbscript however: with most field numbers (the field number is set by the user according to the expected number of colors to be generated) it leaves stripes in the image. Finding a number of fields with which it doesn't is a bit of a lucky strike. In this case the color shift was by 1 degree, which should fill 360 fields exactly. The stripes only disappeared with 363 fields, meaning there is some overlap of top and bottom color (the script keeps pumping out spectra as long as there are fields to be filled).

Even weirder: this picture was created by setting the start color to green, but with output colors inverted. When I set the start color to red, with no inversion, but with the same 363 field count, the stripes are there again. Which suggests that they might be produced by something other than field height rounding. And of course, as usual, the resized JPEG shown on this page will probably display a ton of stripes, but of a different origin. A PNG should never show them, unless they are a part of the image itself, or the image viewer is indeed of a very crappy quality. By my knowledge that is..

Also: the lack of a proper yellow does annoy me. The algorithm plays safe, and in doing so, ends up having no personality whatsoever. How human.