but doesn't this close the door to images with colour gradients?
First, and take this with a pinch of salt but I concluded a long time ago that interpolating gradients between calculated points on the Mandelbrot Set was cheating, and unnecessary.
(To me) The interesting bits of the MS, are the details that are too fine to interpolate -- ie. the adjacent individual pixels all have different values -- not the interstigal bits which tend to fade away to blandness at higher rep counts and resolutions. As such interpolating those bland spaces detracts from the good bits, and is really only used to distract from the low resolution of the pretty bits.
That said, interpolation does make for some very pretty pictures without having to wait hours for the render.
And good news: no, the run length encoding scheme doesn't mean you have to give it up. In fact, it has done half the work you would need to do anyway.
Given the run sequence (count/color): [10,100], [10,200], [10,300] then you want to get to a pixel color sequence something like:
100, 107, 113, 120, 127, 133, 140, 147, 153, 160, 167, 173, 180, 187,
+193, 200, 207, 213, 200, 227, 233, 240, 247, 253, 260, 267, 273, 280,
+ 287, 293, 300
Note: Those are really pre color mapping iteration counts rather than colors.
Of course, if your graphics library doesn't provide an interpolate line draw, then you're back to drawing individual pixels; but the run-length encoding tells you how many pixels and between what range of "colors" you need to interpolate.
And if your GL does do interpolated line drawing then you have the information to set the color end points and lines lengths to hand.
With the rise and rise of 'Social' network sites: 'Computers are making people easier to use everyday'
Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error.
|