Overstepping in v1.67?

[Madman]

Hi Jesse,

I came across an interesting difference between v1.66 and v1.67 when I reopened an old file of mine (Just wanted to see if there was something in it for future use). It's a quite simple hybrid of a Mandelbulb^8 and a Menger sponge. It didn't render quite as nice in 1.67 as in 1.66 with the default settings. Decreasing the Raystep multiplier helps, but doesn't completely solve the problem. To show you the problem, I've attached a preview render of v1.66, v1.67 and v1.67 with a reduced raystep multiplier to 0.1. Am I doing something wrong or is there a small bug in v1.67? I can post the parameter file if you need it.

Regards,

Marco

[Jesse]

Hi, this is a case where you have to decrease the Raystep multiplier a bit further.
Sometimes the new version is better, maybe a bit more often the old one.  But in average the speed advantage shifts the benefit to the new version, also with decreased multiplier.

You can send me the parameters of course, so i have one more testparameter if i will change this behaviour in future.

[Madman]

OK thanks for the quick reply Jesse. So it's been a deliberate trade off between speed and accuracy. Just wanted to make sure that it wasn't a sneaky bug

Parameters below:

Mandelbulb3Dv16{
J.....S....O/...w.........UcH51HzPcozmS.LwL2X5.E3htbO3Awkvv5v0QaDbPozuglrL.thGwj
................................dkKDsB7Gz.2........Y./.................U9/2.....
.................ME//....6Ed0...J0....E......cwAHgffV8mD/..........c./...w1.BnAH
z.....kD..../..........wz.......................................................
........XSlUtuOcWw1..........CskbdRd4uYjjZ.IjzXX/tXdCvqGojb5zoANidE7iJlDxgFDIzDy
6tXrOe11ZsK3zyT8UX2xvtlD......oP.........................2kz....................
.............................UTRB9.Wqpg.MQLn0UuRB9.Cspg.6XLn0UJSB9..............
...........................U..6.X....M4...EB....W....61...EE....4/...I1....UJF1.
...U.y7lzzz98RrzTGwzzzWGr/6.mc..U1Cs.El18XGQeGyDjvIRhrVAkz1............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....6....E..........6....EGiFLNbJaQU.pPrJaQ.........................c..........
...................6./........zj................................................
................................................................................
.....................kG...U5....8....oINiRKNmB1.fZqA.MoI........................
.....MU/4MU/4........cNaNaNaNK.E........kz1........wz.........zD................
................................................................................
................................}

[Jesse]

The previous version has a kind of regulation term, so you didnt had to vary the Raystep multiplier so much.
And on some certain cases like this, it also does a good job on not overstepping, in other cases not.

Maybe i can include a regulation term again that does not slows down the general calculation, however these parameters are good for this issue - thanks!

[MarkJayBee]

Some excellent and useful new features in version 1.68.
Unfortunately though, I've been having the same problem with
overstepping as Madman.
In my case it presents as black, noisy pixels, and seems
to occur in many of my images since I started using v1.67.
v1.66 and prior never exhibited this problem.
I've tried reducing the RM to 0.01 and the noise is still there.
And typically on images that really need to be clean to work -
isn't that always the way?!
Also noticed a lot more clipping when navigating in this version too,
keep ending up stuck inside objects! 

If anyone has a suggestion on how to obviate this problem - I'm all ears!

I'll probably just wait and hope Jesse includes this 'Regulation term' in
the next version.

Attached a couple of examples:

[Jesse]

Mark, there are formula combinations that just wont work because of bad DE's.
Thats independent of the program version, but because of more different formulas there are also more 'illy' combinations.

If your examples worked in version 1.66, please give the parameters and i will try to figure out what the problem is.
Could be another problem beside the regulation term.

PS:
Sometimes if you have steps in the DE, the normals are corrupted.  Then you might uncheck the 'Normals on DE' option,
but this will only work for some formulas.

[MarkJayBee]

@Jesse, thanks for the quick response.
I did a render of the params in v1.66 and the problem exists there too,
although not to such a severe degree.
I suspect that I've found an 'illy' (iffy?) formula that M3D just doesn't 'like'!
I've included params for both versions below:

v1.66:

Mandelbulb3Dv16{
J.....u1...o5...w.....E...UUjP6ft/n2.fSj6//Zqi1EkZ7jA.Ny0zvkLMKDHbUrzq82uqrL97.k
................................ZDTRtx2oqz1........Y./..................y.2.....
...................2/....6UqA...H1...2E4.....6sbArLq4WnD/..........c./...w1.BnAH
z..kviPD....1..........wz.......................................................
........6xEnOBUvYxXeRREjX7.WyYtZ2m4qBppDDffFZjH7/rP.ILCsrlMOzotApktwhHdj88VyiVRH
RxvVNMe5uHGYywCe9Yp.iHqD......II54....................sD.2kz....................
..............................nxF5.kr5R.EVTo/.CyF5.Qt5R..cTo/.dyF5..............
...........................U..6.51...60...EB....//...A0....F....8/...s04...UJF1.
...U.0aZczDM6/nz2tIRzHfiM16.mc..0HPV.MxuvBTuCfxDU1Dbqxm4Iz1..........2aA8.kzRf8.
EZqulQA8iz17H0/bvOozz.........../6U0.wzzz1................................E.0c..
zzzz.................................2U.8.kzzzD.................................
/6U0.wzzz1...................................APYW/kgF84.zzVhtiA..1Ak.szDoZ5h.Mtf
f1EuOhdht0kaqa9.ftrgF84.XNMg..bTsLM5.UTVR..lyVTVR..y3q/.Mw5y3q/.sLM5.kqTsLM5.UTV
R..kzVTVR..y3q/....y3q/.zz/k.1A.yz1yAT2.xzpaqa9.................................
E....6....E.....W....c....kIdJaQkZaPnhKOn.UQ..............................U/4MU/
4M.................../........zD........kz1........wz...........................
................................................................................
.....................2....E9.........wZQZBKOk7rPX34PM/..........................
..............................0E................................................
................................................................................
................................}


and v1.68:

Mandelbulb3Dv16{
K.....u1...o5...w....2E...UUjP6ft/n2.fSj6//Zqi1EkZ7jA.Ny0zvkLMKDHbUrzq82uqrL97.k
................................ZDTRtx2oqz1........Y./..................y.2.....
...................2/....6Un6...A1...2E4.....6sbArLq4WnD/..........c./...w1.BnAH
z..kviPD....1..........wz.......................................................
........6xEnOBUvYxXeRREjX7.WyYtZ2m4qBppDDffFZjH7/rP.ILCsrlMOzotApktwhHdj88VyiVRH
RxvVNMe5uHGYywCe9Yp.iHqD......Y0a3....................sD.2kz0...................
.............................UWRa4.iqNO.6Rbd/UxRa4.OsNO.sXbd/UMSa4..............
...........................U..6.51...60...EB....U0...A0....F....8/...sm6...UJF1U
...U.0aZczDM6/nz2tIRzHfiM16.mc..zrhe.MxuvBTuCfxDU1Dbqxm4Iz1...........a.8.kzRf8.
NqRaqWXXzz9osKtn7mv/./........../6U0.wzzz1................................E.0c..
zzzz.................................2U.8.kzzzD.................................
/6U0.wzzz1...................................APYW/kgF84.zzVhtiA..1Ak.szDoZ5h.Mtf
f1EuOhdht0kaqa9.ftrgF84.XNMg..bTsLM5.UTVR..lyVTVR..y3q/.Mw5y3q/.sLM5.kqTsLM5.UTV
R..kzVTVR..y3q/....y3q/.zz/k.1A.yz1yAT2.xzpaqa9.................................
E....6....E.....I....c....kIdJaQkZaPnhKOn.UQ..............................U/4MU/
4M.................../........zD........kz1........wz...........................
................................................................................
.....................2.....3.........wZQZBKOk7rPX34PM/..........................
..............................0E................................................
................................................................................
................................}


Cheers!

[Jesse]

yes, i already thought of _reciprocalX 

that formula makes the problems, of course.  Was thinking of including it or not, it was a suggestion by bib and i know that he can do things others can't... so my suggestion: usage of this formula only by bib and the really tough ones 

[bib]

he can do things others can't...

Like biting his toenails or touching his nose with his tongue?

[MarkJayBee]

I perform both of those feats as a matter of routine! 

@bib, Would you take pity on a poor old M3D addict, and share your
method for getting rid of the noise when using ReciprocalX?

Think I've seen a couple of your pieces which look like you've
used that formula. Did notice the colour scheme was quite dark
in both - does that work to hide the noise?

[bib]

I have no particular trick unfortunately. I just try to play with the camera position and various parameters (max Stepcount for binary search), and especially as you noticed with coloring and fogging (low Dyn. Fog and max ambient shadows)