Welcome to Fractal Forums

(http://mathr.co.uk/mandelbrot/2013-06-14_emndl_perturbation_test.jpg)

http://archive.org/details/emndl_perturbation_test_2013-06-14

Test render using perturbation method (some of the maths from superfractalthing), with glitches around deep islands and embedded Julia sets - I have some code that tries to detect where the delta method went wrong, but it's not finished, and when it is I'll still need to add more code to re-render the bad parts with more appropriate reference points...  Source code is here in the 'perturbation' branch, but it's fairly hideous/messy at the moment:
https://gitorious.org/maximus/emndl
http://code.mathr.co.uk/emndl

Cool, I like your rendering "lines". It would be really cool if you are able to detect and correct the glitches, I solve this problem by not going too close minibrots on the way to the final minibrot ;)

Nice video
You've got an interesting rendering method, too bad there are those glitches, and I like the music too.

The lines are from using distance estimate for colouring, in HSV space hue = continuous iteration count / 512, saturation = 0.5, value = tanh(distance estimate / pixel size), with non-escaped points coloured separately.

I like going close to lots of minibrots, makes more interesting shapes stack up :)

Thanks, I'll try and fix the glitches later this week - the approach I've been working towards so far is:

• detect badness by seeing if neighbouring pixels have identical continuous iteration count
• get rid of tiny regions of badness that might be false positives, and expand larger regions into smooth blobs
• recalculate each of those blobs using a better reference point
• repeat until no more badness

http://mathr.co.uk/misc/2013-06-08_detecting_bad_references.png shows the first couple of steps.

The music isn't mine, it's just something I found online that was the right length.