Welcome to Fractal Forums

OK, this is really nerdy... But at least I find it interesting  O0 I hope someone can answer...
I was comparing the render I did to celebrate 2 years with the bulb:
http://www.fractalforums.com/index.php?action=gallery;sa=view;id=9294
With Daniel Whites wonderful "Cave of Lost Secrets":
http://mandelbulbs.s3.amazonaws.com/new/full/q50/MysteryCave-b-med.jpg

And because I have done hundreds of renders of the different mandelbulb versions I can tell that both PROBABLY are Cosine bulbs (because the SIN version are much more symmetric) BUT there is a small difference between the two images, you see the "stay" in Daniels render? In mine there is nothing like it... What could make a small difference like that? I know I will probably get no answer :) But I hope someone can give me a hint...  O0

Mmmmh, isn't it again a matter of turning the image upside down? (see attachment)

By the way, it's not a stay, it's a wing:

Strange, I get no Wing... not anywhere... is this MB3D fresh renders of a cosine bulb bib??
And thank you for bothering! :)  :beer:

Yep, fresh CosinePow8


Mandelbulb3Dv17{
T.....S....O/...w....2.....lUQQAHfflzma4FmWnIE/EB39gsP6i2znrW1nZKvktzk/7z8b16JwD
................................UYi/5psFb.2........A./.................U9/2.....
................/M.0/....6Ef....q.....E2.....2huIXca7YnD/..........m/dkpXm1....U
z.....kD12../..........wz..................................u1....y1...sD...../..
.w1...sDG.5H5pmyswXYjHZmQyYxyqTipWZyfriDU4jhi1Oxgv9sCr0ZfNTsy6tWnsNn1YnDOldErx2c
qvPejive8sDCz03VcqF9D/iD......o3..............kD.2....sD..kz....................
.............oAnAr1...sD....zUJRVA.yp3m.MOL61UnRVA.qr3m.sVL61UFSVA..............
..................kz.wzzz1.U..6.t.....7....F.........M0...U8....8/...I1.....SlS0
...U.qFG9yzb2zzzRYoWzz7lz16.0c..zXCc.YHbG7eFT5uDKjySnbiney9..........2.28.kFrA0.
FWb96aAIVzvh1se7Umvxz0........../6U0.wzzz1................................E.0c..
zzzz.................................2U.8.kzzzD.................................
/6U0.wzzz1...................................2CcN/UvNP6.eeWCNq0.yRii.EJJUk1f..XR
S1.mx3CcN/UvNP6.QsLsUa3.ibhV..bTV1OK.sSq40.ly3CcN/UvNP6.MwLsUa3.ibhV.kqTV1OK.sSq
40.kz3CcN/UvNP6...EsUa3.eeWCNq0.IJ36wk8.wyLsUa3.................................
E....2....E.....I....2....kEjBLOiJ4IjR5C.IaQ....................................
...................wz.........zj................................................
................................................................................
........................}

OK, I will try your parameters bib... If I turn mine upside down it does not match Daniels at all at the top... the whipped cream should be up...
Holy... is this geeky or what? :)

Now, this is really weird... rendered your params... Is it my 'puter????  :-\

I put in your render for direct compare:
(http://www.fractalforums.com/index.php?action=dlattach;topic=9439.0;attach=5940;image)

 :hmh: :hmh: :hmh: :hmh: :hmh:

Jesse will explain that. Sounds absolutely mysterious. What OS are you running? I'm on W7

XP (On Parallels on Mac OS X 10.6.8 ) ...might be the "problem"?
Strange it only affects the "wing"

BTW, I like your image of the wing very much 8) You should do a high res render of it!!!
(http://www.fractalforums.com/index.php?action=dlattach;topic=9439.0;attach=5941;image)

Of course, i will explain it, not today but certainly somewhere in the future.   :dink:
(Because bib could foresee it)

If it was a win issue i would say that Bill G. would have needed the wings to meet Benoit M. for some pints of beer...
but since it seems to be a Mac emulation issue and i don't think that Steve J. needs kind of... i really have no clue!

I would request some other Mac Parallels images to see if it belongs to this combination, btu if only belongs to your
personal computer, KRAFTWERK, i would be scared!

Hope this helped  :dink:  :beer: :beer: :beer:

Jesse I am experiencing issues with pasted params too :)
Some pars don't render the same in all computers it seems

Ok, but this is very general spoken.. it depends on the program and formula versions involved.
And since you are updating your formulas quite often, you are responsible for at least half these issues  :dink:

But this special issue of wings (missing) is really strange because i don't think i have changed the CosinePow8
at all, and it should not affect the program version too.

But maybe this is a kind of conspiracy we run into, who knows..

Never mind, found the cause , they did not update formulas :)

Very mysterious "answer" Jesse... ;)
What images do you need to solve this?

Maybe this topic should be moved to the Mandelbulb 3D forum...

Darkbeam is probably right, as usual in IT, a problem is a matter of component version. Grrrr, he broke the charm! I liked this idea of the mysterious wing! Anyway this does not give the ultimate answer: is that REALLY a formula version issue, and if yes, what changed and why only this little impact? I would be curious to see the corresponding "artifact" on the Power2.

Jeremie I was speaking for my formulas and as Jesse said CosinePow8 is unchanged for a long time (the charm remains O0 ) and the thread has been moved by a psychocinetic energy :dink:

hey people

2 things:

1st: i believe the perspective isnt exactly the one from daniel, it looks a bit rotated, i mean the viewpoint may be the same, but camera position certainly isnt ...
2nd: the different renderings come from:
        a. a general broken load/save functionality, this could be checked against that the user which created the image (bib) uses his own params to rerender the image after a new startup of the program
        b. different formula version

;)

A mysterious answer to a mysterious topic... should fit  :dink:


Just wanted to say that anybody with a system like yours should proof bib's parameters if they get the same result as you, but anyways, i would have still no clue what this strange thing could cause.  Some inaccuracies in specific calculations???

Thank you for your concerns in this (rather interesting but geeky) problem of mine...
Christian: At the viewpoint of bibs render you clearly see the "wings" and my render of bibs params are from the exact same angle, even if there seems to be a slight difference in the FOV... hmmm, do you see this too bib & Jesse... My "camera" seems to have a wider FOV and "looking down" a bit more... was it your exact parameters bib?
Maybe if you save the parameters as a m3p file instead bib???

I will also make sure I use the latest formulas (I also thought that the Cosine bulb didn't change since the day the formula was found?)
But I have been slow on changing the ini dirs... I will try to do that this evening.

And OK Jesse I get what you were after. Well might be better not to fix this... I might never find the Wasp Troll again :)

No. In fact I did not save the params for the 1st picture. Then you asked for the params so I tried to do approximately the same image, but obviously the PoV is not 100% matching.

OK, then this is not an issue... I will try to refresh the ini dirs and see if that helps. Thank you bib!

With the same params it's weird to notice the difference in quality and very slight difference in PoV
Your image:
(http://www.fractalforums.com/index.php?action=dlattach;topic=9439.0;attach=5943;image)
Mine:

Still a slight miss-match...  O0
And my image is a bit more noisy... Strange...

have you ever considered something similar to the "pentium bug"? maybe bib is running an amd system...  :hrmm:
 :rotfl:

Yes why not  O0... are you on AMD bib?

Intel core i5 (my laptop)

Have not tried on my desktop (now on Ubuntu + Wine, Intel core i7)

Same processor here!

I get the same picture as bib... Win 7 64 bits + i5. Since that's 2 out of 3, Kraftwerk must be wrong, statistically speaking   :D :beer:

This looks like a problem with detecting when a ray intersects the surface to be rendered. If a feature is thinner than the sampling step it may be missed entirely. Where sampling starts is usually depends on the view point. Some rendering algs will bias their search near where neighboring pixel rays intersected the surface. This causes large features to appear and disappear based on small changes in the viewpoint.

I agree in theory, but in practice, even by decreasing the calc precision to try to step over the wings, they remain. There must be something really weird happening on Kraftwerk's machine.


Mandelbulb3Dv17{
T.....S....O/...I....2.....lUQQAHfflzma4FmWnIE/EB39gsP6i2znrW1nZKvktzk/7z8b16JwD
................................UYi/5psFb.2........A./.................U9/2.....
................/M../....6kM....C/....E2.....2huIXca7YnD/..........m/dkpXm1.BnAn
x.....sD12../..........wz..................................u1....y1...sD...../..
.w1...GEG.5H5pmyswXYjHZmQyYxyqTipWZyfriDU4jhi1Oxgv9sCr0ZfNTsy6tWnsNn1YnDOldErx2c
qvPejive8sDCz03VcqF9D/iD......I2..............kD.2....sD..kz....................
.............oAnAr1...sD....zUJR0B.yp7o.MObE1UnR0B.qr7o.sVbE1UFS0B..............
..................kz.wzzz1.U..6.t.....7....F.........M0...U8....8/...I1.....SlS0
...U.qFG9yzb2zzzRYoWzz7lz16.0c..zXCc.YHbG7eFT5uDKjySnbiney9..........2.28.kFrA0.
FWb96aAIVzvh1se7Umvxz0........../6U0.wzzz1................................E.0c..
zzzz.................................2U.8.kzzzD.................................
/6U0.wzzz1...................................2CcN/UvNP6.eeWCNq0.yRii.EJJUk1f..XR
S1.mx3CcN/UvNP6.QsLsUa3.ibhV..bTV1OK.sSq40.ly3CcN/UvNP6.MwLsUa3.ibhV.kqTV1OK.sSq
40.kz3CcN/UvNP6...EsUa3.eeWCNq0.IJ36wk8.wyLsUa3.................................
E....2....E.....I....2....kEjBLOiJ4IjR5C.IaQ....................................
...................wz.........zj................................................
................................................................................
........................}

Bummer.

Uninformed speculation B.

The wingless render seems to have a higher level of detail (or more noise as Kraftwerk points out). This is often associated with a higher bailout radius or higher iteration limit. Unfortunately I cannot read m3pese, could this be an issue.

Wild speculation C

It was not clear to me from reading the thread, but is the only difference the processor type? Another source of difference is optimazitation level. Depending on the compile switches, the executable may pick the code to run based on the processor type. Normally this does not matter, but iterative calculations can sometimes magnify precision differences.

Thanks again for your interest in my lack of wings!  O0

I think the bailout level should be included in the parameters... could be wrong... Jesse? Bib? Well I can check it out myself too. :)

Do not know anything about compile switches and optimazitation level, I guess Jesse needs to answer that one.
But we have the same processor type, the difference is that bib runs a pure Windows 7 and I run XP (On Parallels on Mac OS X 10.6.8 ).

I attached bibs parameters as m3p file, so we can exclude a disfunction on the text parameter loading.

This must be a bigger change of values, not some little inaccuracies.  It is not a big zoom-in and only a few iterations, even single precision should be enough to get this detail!

You might reload the actual formulas to make sure nothing was changed in there.

ps: changing the z multiplier to 0.9 makes the wings smaller too, but changes also other structs...

I'm on an AMD machine and I get the wings... :)

Thank you Jesse, I will try this m3p with reloaded formulas this evening!
I did not change the Z multiplier...

You could do commercials for Red Bull !  O0

BTW, Hal, did you use bib:s pasted parameters or Jesses m3p-file?
If bib:s we can rule out problems with pasted parameters?

I don't think there's anything wrong with the copy-pasted text parameters form bib. When I paste them in M3D, I get the same "view" as KW, but the "smoothness" from bib's picture. I'll add so you can compare it yourself...

And i tried to reload the cosine p8 formula + use the m3p file from Jesse - no wings.  :angry:
Well, I guess it must have to do with XP / Parallels / OS X somehow...
I will try to get Windoze 7 installed on VMware to see if it makes a difference...

XP as only reason i can exclude. on a native XP SP3 i render Jesse's m3p with wings.
maybe it is however a spell or a curse :yes:

:) Probably the Wasp Troll....  O0
But might be parallels, so I will test VMware with W7

I have the only possible solution:

You have changed some of the "Constants" values in the formula and forgot to rechange them or didn't remembered the correct ones:
1e-40, 7, 8, 28, 70  are the correct values.

But you have to "Remove this formula from the local appdata folder" by rightclicking on the formula in the dropdown formula tab first, so the invalid one can be replaced with the original one.

Maybe you got a 27 instead of the 28 there... must be the reason!!  (else you are cursed, that is for sure  :dink:)
 :beer:

BINGO! You are a genious Jesse, I have no recollection what so ever of changing something, and don't know what values I had but I did as you said and now I*VE GOT WINGS!!!   :beer:O0

You are a spaghetti-formula mangler! (almost like me!) Congrats. :tease: :evil1: :dink:

I guess that :beer: was actually Red Bull ;)

Must have been! ;)


More of a macaroni-formula mangler I'm afraid...  O0 Thank you Luca!