Welcome to Fractal Forums

Hello ;D
I have some question about the interpolations in mandelbulb3d
Sometimes when I change some parameters between two keyframes the animation plays the transition sometimes not.
What are the parameters that can change by always keeping the interpolation between two fractal.
I try to create transitions between two different formulas but I can not

thank you !!

You wont see results using two different formulas.  Animation is done by changing parameters within a single formula (or set of hybridized formulas) but there is no way any fractal generator can animate between 2 entirely different formulas.

Hope that helps.

Of course you can morph between two formulas using interpolation and animation sockratease.  :dink:

Perhaps I misunderstood the question - I meant, for example, morphing from a mandelbox to a mandelbulb.  That wont work.

Changing parameters on either a box or a bulb will work, and indeed is how we all do things, but that is all with the same formula.

Changing from one formula to another, like box to bulb, will never work!

Mmmmh I have just been morphing between a mandelbulb and a dodecahedronIFS...

You select interpolate and two formulas and then you can create a morphing animation with keyframe one having weight zero for formula one and weight non-zero for formula two and keyframe two having weight zero for formula two and weight non-zero for formula one or vice-versa.

It's not perfect though as there is a too abrupt transition between 99% dodeca/1% bulb and 100% dodeca/0% bulb, which is perhaps related to what the OP speaks about, but other than that it works (obviously) and is pretty cool.

this is wrong, especially when using interpolated hybrid, a transform from one fractal to the other can be achieved by simply blend between 0 and 1 of the interpolation.... and this is possible within mandelbulb3d

I thought the abrupt transitions were unavoidable, and in every case I tried they were so abrupt as to seem like the whole thing was not possible.

Always glad to be wrong when thinking something can't be done though!

I'll have to try it again and see how it works...

it does work, but only when interpolating the values ( at best using spherical interpolation of the triplexes ) at each iteration, and the interpolation rule states that

if:

f1 is fractal1
f2 is fractal2
t is the interpolation value 0..1
then (e.g. linear-) interpolation with t=0 would result in fractal1 and if t=1 then fractal2 is the result, the interseting stuff happens in between, i have done a ultrafractal5 formula, which allows to blend virtually any fractal with same bailout logik... even a transform of julia to mandelbrot would be possible, i render a test right now...

use the blend parameter of the formula to blend between 2 fractals...


Fractal1 {
fractal:
  title="Fractal1" width=640 height=480 layers=1
  credits="ckleinhuix;1/20/2012"
layer:
  caption="Background" opacity=100
mapping:
  center=-0.0875/0.0625 magn=1
formula:
  maxiter=250 filename="Standard.ufm" entry="GenericFormula"
  p_formulaClass="ck.ulb:BlendFormula" p_formulaClass.v_generic=100
  p_formulaClass.v_formula=100 p_formulaClass.p_power=2/0
  p_formulaClass.formulaClass1="Standard.ulb:Standard_Mandelbrot"
  p_formulaClass.formulaClass1.v_generic=100
  p_formulaClass.formulaClass1.v_formula=100
  p_formulaClass.formulaClass1.v_divergentformula=100
  p_formulaClass.formulaClass1.start=0/0
  p_formulaClass.formulaClass1.p_power=2/0
  p_formulaClass.formulaClass1.p_bailout=4
  p_formulaClass.formulaClass2="Standard.ulb:Standard_Julia"
  p_formulaClass.formulaClass2.v_generic=100
  p_formulaClass.formulaClass2.v_formula=100
  p_formulaClass.formulaClass2.v_divergentformula=100
  p_formulaClass.formulaClass2.seed=-0.20238/0.65476
  p_formulaClass.formulaClass2.p_power=2/0
  p_formulaClass.formulaClass2.p_bailout=4.0
  p_formulaClass.interpolator="ck.ulb:ComplexInterpolator"
  p_formulaClass.interpolator.v_generic=100 p_formulaClass.blend=0
inside:
  transfer=none
outside:
  transfer=linear filename="Standard.ucl" entry="Smooth" p_power=2/0
  p_bailout=128.0
gradient:
  smooth=yes rotation=1 index=0 color=6555392 index=64 color=13331232
  index=168 color=16777197 index=257 color=43775 index=343 color=512
opacity:
  smooth=no index=0 opacity=255
}

The usage of the interpolation hybrid to blend between formulas depends much on those formulas, not every combination works really good.
In m3d both formulas uses the interpolated vector as new input, maybe it would give better morphs to just interpolate for bailout check and for DE calculation in the end?

But i could think of even better ways for morphing, like a somehow weighted version of the smooth DE combination.
Dunno exactly how this must be implemented, but it would be cool if one fractal would kind of liquifying and morphs into the second one that gets solid... just a thought.

Blimey!   :elvis:

It Works!   :clown:

(http://www.sockrateaze.com/stuff/proof.gif)

I had to really tweak the keyframes.  It seems going all the way to 1 causes a sudden jerky transition no matter how many sub frames you use - but if you keyframe the weights something like {1,0 : 1,1 : 0.1,0.9 : 0.001, 0.999} you can get smoother transitions.

So there's a Mandelbulb morphing to a Menger Sponge!  One of the sort I would have thought impossible just this morning...

 I attached the m3a file if anyone wants to see how it's done, despite it being obvious in retrospect.

Just goes to show.   :music:

as i said ;)

thank you very much


it can only be interpolated with the formula hybrid "interpolate"? I try to do with "alternate" and I can not, is there a solution?

thank you again

when using alternation, you create new formulas, and create new parameters, but you can not go back to the basic formulas,
when using interpolation, you can blend between two independent formulas....

Unless you do alternated blending - or even blending across iterations ?

sure, you can combine everything with everything, so, you could blend to alternating formulas together ... that should lead to the blend effect,

i just say:

when iteraten 2 functions parallel, f1 and f2
if you use linear interpolation, and interpolation value is 0,
at each iteration step exactly lerp(0,f1,f2) -> f1 is the result,and lerp or better slerp (spherical linear interpolation )
leads to a new formula, as sock demonstrated it is working on certain fractal combinations, you have to combine diverging formulas together, or converging formulas, but not mixing those two ...

people, i am drunk, did i mention i love this forum ? considering i am just moderating it, it is a real pleasure to see it thrive even more....

Hi again,

Gosh folks -  I think that this interpolation between two formulas is just what I was hitting on at the bottom of my lengthy post as a newbie @: http://www.fractalforums.com/mandelbulb-3d/how-to-create-%27unique%27-fractal-art-with-m3d-%28or-how-not-to%29/msg33594/#msg33594 - but at that time I did not know how to deal with some of the bizarre effects of doing so - like the Navi not showing the same image that had been calculated or how to get the M3i to recreate what was between the anim keyframes - and - it seemed at the time, no one was really able to offer a real solution or couldn't understand what I was trying to say. Maybe I didn't explain myself very well at that time. It felt like I was waffling nonsense or using the program 'wrong' sometimes. It seems from this thread that some some of this is better known now - at least to some. I wonder if this was always the case or if my post has helped others try this method out (?)

Anyway, if it helps, I have learned a few ways of improving matters since then:

One is to make sure that the opening program default parameters are gotten rid of by sending an image from the anim keyframes into the main window. (You don't have to render it)
two: simple use of the 'M' below the Calculation box often (but not always) works to create an M3i file that keeps whatever weirdness has been created. It took me ages to spot this one.
3: If loading the M3i back into the same open program still does not load the correct picture, try loading it into a completely new instance of the program. This sometimes works. Make sure you have saved the correct colour scheme as this is often completely wrong on reload (which can also work out ok too).
4: If all else fails and the massive jumps between changes cannot be overcome or you cannot get the right params into the Navi to gain two new keyframes to render between - (sometimes  - not always) a way around the jumps at least is to make massive (and I mean like 10,000 or 20,000 ++?) subframes between the two rogue keyframes - but you only need to start rendering from the number where the change point you want starts to occur (trial and error - it may be from, say, 9999 to 15999) and, with luck, you will generate sufficient subframes to give a smooth animation of sufficient length, gaining pictures you (or at least - I) cannot seem get any other way (yet).

Hope this helps. I didn't post these new tips as an addition to my original post because I thought it would still appear like garbled nonsense to folk. Hopefully not now.

regards
Kam