Welcome to Fractal Forums

I'm not sure if if it has something to do with the hybrids I'm creating lately or if it is an issue with the latest M3D version, but using hard shadows seem to create additional noise. The effect seems strongest when using hybrids with Folding Integer Power as one of the formulas: Smooth without hard shadows and noisy and grainy with hard shadows. Has anybody else seen similar effects?

I haven't noticed any increase in graininess.

There is a little issue that showes up only in certain fractals, generating some little black streaks to the right side of shadows...
this will be fixed in the next version.  I guess this is due to the fact that the hard shadows takes account of the variable DEstop on Z, since version 1.697.  Some speedup functions seems to have trouble with it.

Think I've used Hard Shadows maybe twice since I started with MB3d, to lazy to get them set right and too impatient to wait on the additional render time. I'd have definitely been yellin if FoldingIntPower was afflicted with noise otherwise though, use it all the time.

Agreed, there are certain types of hybrids where hard shadows have a disastrous noisy effect all over the surface.

Thanks. I haven't seen any black streaks like Jesse describes. It's more like bib's observation: Surfaces that seem smooth at first have a mottled look after Hard shadow calculation. Playing with DE-stop seems to help a little in some cases, but any of the other noise reducing techniques have little or no effect.

I guess we'll just have to wait for the next version with noise reduction AND reflections!  No pressure intended, Jesse  :dink: ;D :dink:

Could someone please post parameters with this noise issue?
 ...just to be sure that i solved it in the next version!

The reflections are making some progress too, after a little hardware breakdown.
Running on half memory now, hope that no breakdowns are happening anymore  :-\

Dunno if this is the kind of thing you're talking about madman, but
I've noticed this 'leopard skin' pattern - as bib christened it! - in some
of my renders which have predominantly smooth surfaces. It's acceptable
in some images, but I wouldn't want it every time I use hard shadows!

(http://nocache-nocookies.digitalgott.com/gallery/7/2025_11_05_11_5_29_37.jpeg)

If the the params would help, let me know?

Thanks Mark, but that soft spots are more due to the ambient shadow on DE's, i guess.
Because the DE's with reciprocal formulas vary much.

I discovered once some HS noise, but more test images would be better, so...
post also directly to me if you don't want to publish the parameters, thx!

I usually do not save faulty params but if I encounter it I will post it.

Btw the reflections seem absolutely fantastic, a true game changer, again!

On the Reflections *note*
I'm very familiar with raytrace Reflects/Refracts from using MojoWorld and Bryce. In those apps heavy Reflects added exponentially to Render times, will it be the same for MB3d?

OK, time to post some samples then. See the attachments. Both are details of a larger render. The first one is with hard shadows and shows the grainy look and the second one is without hard shadows. These are the parameters:

Mandelbulb3Dv16{
O.....Y/..k72...6....26...kmrRmk7tuizk4pXNuLIB/EsVi.RjDtUz9M7FT/huPozk9davf9F/wD
................................S735WWGpc/2........Y./..................Y.2.....
................/s10/....6Ev7...B/...2EC.....2K.//QRekhD/s1........c./...w1....U
./EnAnQD/...02EmlVuauz.3.bwuGSZKu2.kYjx/fWyA3.2...................2...sD...../..
.w1...sDKgOL1FPihuv9DXl7jhyqy6DA67ttKIYjNVcHiZ54qtfGw7fml4U2y8OqorkudkhDgvqGzIiu
PvHBor6aecPfyktY8a7WkLbD......Yd2...D4................sD..E.0...................
..............................MRS3.4qtJ.kObL/UoRS3.srtJ.sVbL/.FSS3..............
.....................wzzz1.U..6.P....61...EB....I....U3...U2....F/...I1....UJl22
...U.ydelyjeYFnzTeOgzf8No.6/0EVzRHLI.QvubxYsDCtDwGGmc8YlJz1W.o0aMg.jz.U.8.U4Ux..
IwUmc2beYz1RdA8E5Exwz0...........6U0.syy61EXrpgPsZKyz0......................0c..
O5yw.oklrlkCx9zD5Got8VlJqy9..........2U.8.kzzzD.................................
/6U0.wzzz1....................................uBZ.EHVM/.zzFoTu2.cxo7.szDnueM.g5P
D/EzTt6Ul/kF.V1.Qs5crI0.B3W3..bTUSH7.oI6K..ly/uBZ.EHVM/.Mw5crI0.B3W3.kqTUSH7.oI6
K..kz/uBZ.EHVM/....crI0.zzFoTu2.yzngi84.xzZX.45.................................
E....6...../....I....M....EE0x4SK3aQtBpMVlKNoE4.......................k/..E.....
..........EnAnAnAnAzz.........yD........kzXwGrYMEUJmz.........zD................
................................................................................
.....................2....U/....1....MoPgFKOiR467t4RU.pPr/......................
8..0...........................E........d/A........wz...........................
................................................................................
................................}

You know I think I have seen that before, on a couple from a ways back, I know I didn't use HS but I think I used DEAO. If I remember right what I did was up the Max iterations, which of course affected color and the scene a bit but I went with it.

@Madman:  Thank you, the parameters seems to be very good for a test.

@bib:  I would not say it is a game changer, but a nice effect that fits with some images, mostly with many flat surfaces and no great details.
And no urgent need to post more parameters yet, thx.

@lenord:  This effect is not exponential scaled (for time consumption) in reflection depth, because it calculates only straight reflections (and some DEAO rays) and not do a full monte carlo simulation.  That would take maybe some month to render an image noisefree :)

cheers  :beer:

Jesse
"@lenord:  This effect is not exponential scaled (for time consumption) in reflection depth, because it calculates only straight reflections (and some DEAO rays) and not do a full monte carlo simulation.  That would take maybe some month to render an image noisefree"

That's good, perfect. The reflects/refracts in MoJo, although superb. are the Full Monty as you say and render time is unbearably long, 2 days for a 2kX1k is a bit much no matter how good it is.

not really, the first perspective 3d mandelbulb images done by me in 2007 (http://www.fractalforums.com/index.php?topic=742.msg4129#msg4129) had fully unbiased global illumination, and that was before distance estimators were introduced.

i've been wanting to add mandelbulb/box support to indigo renderer for some time, but i'm still unsure how to handle (generate DE functions for) all these new formulae that're going around... because then it's easy to have it working with any kind of 3d scene, any materials, gpu acceleration, motion blur, networking, the whole lot :)

It would be cool to implement those things (Bryce style);

- translucent / semiopaque / mirror materials (I know it's very hard,  :-\ ) - the glow fx is already here (the fog ;) )
- cloud field, moon, sun fakings (this is more fast I think, for the cloud field you can use sfbm + gnarly patterns? :D )

Pwease :'(

 O0

Cool, if not DE then did you used voxel space to render or fixed stepping?

My guess is for some hybrids and more special light options, it would take still a month or more for a fullscreen noisefree render (i am speaking just for me and CPU, of course).


The 4 point method is a very general DE option, but not all formula options gives valid DE's with it.  Don't know how to generate a more general one, beside of analyzing the DE's and make some kind of corrections for it... (just an idea)

All solved in 1.7!! And then you get reflections on top... Just for the fun of it, here's a version of my problematic parameter set in 1.7 . And yes, I had to add some reflections of course ;D

Tks Jesse!