ok, since i use this technique for more than 12 years, i got some insight into it....
first of all, try studying the effects of hybrids in the 2d view
second here are some forms of hybrids, that not all are available, but some forms are available:
i use some descriptions, in this cases formula1 always means standard mandelbrot
alternating - here a new formula is generated by f1(f2(...fn(z)) the result of the one formula is taken as input for the next formula, the effect on creating formulas in this way is that the "iteration bands" are the ones that show the behaviour of a certain formula, e.g. the first iteration is then formula1, for the second iteration ( so to say, the second band ) will then be the domain
of the second formula, and then first formula again takes over, for the alternating of many different formulas you need to increase the overall iteration count to really make use of it
it can not really be intuitive, but somehow effects seem natural, e.g. but it needs to be experimented with
follow ups - here formula1 is used for n iterations then formula2, this is quite easy to control, if formula1 is the mandelbrot, you just place another fractal inside the non-escaped points at that iteration this is quite cool
blending - in this hybrid form 2 formulas are blended over eachother, either by linear interpolating of the 2 values by a constant value ( e.g. 0.5 ) or by spherical/polar interpolating of the results of each formula, this is done for each iteration, when animating the blending value from 0 to 1 it is possible to "morph" virtually any fractal into any other fractal....
optional - in this method, any rule could be applied to differ between the formula to use, best work simple options like x>y, or abs(z)<1, this results in some "folding" operations, known from the barnsley fractal, or best actual example would be the mandelbox ( if constructed as hybrid from its parts )
so, if all would be written in tight mathematical style you would see that each hybrid can perform to a completely new fractal, but when interpreted visually you always will encounter similarities of each part of the generated formula
some hints and tips to try out
- alternating form this is the standard which is available in both renderers, just use 3 alternating formulas of exactly the same type, this would have the effect that it would look exactly like to choosen formula, but with tripled parameter set ... so, try animating wildly the properties of each, keeping in mind that you can easily transform it back to its original state by smooth transition, but it would not have to be a forward-backward animation, i mean the backward animation to its original state can be anything you like
best viewd with 3 keyframes, the first to make an inital deformation, the second to force a different route to the starting point, and the 3rd keyfame for the original position, this would create a smooth loopable animation
so, this is what comes into my mind upon your question, report insights if you find some
and have fun exploring