PP2014, Cellular Node: How to get rid of that godawful repeating pattern ?

  • The Cellular node can be used to create great-looking worn little tiles, except that the goddamn pattern repeats every ten or so cells !

    I know that this repeating of a pattern is a problem with most (all?) of the supposedly 'random' 3d nodes in the material room (Noise, etc). I recall that suggested workarounds included plugging other 'random' nodes into some of the inputs.

    (Also, and slightly off at a tangent, if you want to apply a 3d node to a 2d UV-mapped surface I recall that you can plug (for example) a U node into Scalex, a V node into Scaley, and a Math_Functions node set to output a '1' into ScaleZ.)

    However, try as I might, I cannot seem to get rid of this irritating repeating pattern coming from the cellular node...

    0_1525864437562_Render 1.jpg

    I can't seem to find previous threads I've posted to on this subject - I think they were at RDNA

    (the shader used in the render is from here http://www.sharecg.com/v/83427/gallery/11/Poser/Three-More-Poser-Procedural-Shaders )

  • @3dcheapskate have you tried using the Cycles brick texture?

  • @3dcheapskate I've had similar problems when trying to eliminate regular patterns in a FireFly-only micropolygon displaced grass texture with various scales of tiled, seamless texture overlaid. The human brain-eye combination is exquisitely sensitive to repeating patterns. @ghostship 's suggestion to try Cycles nodes is certainly where I would have started, too. Fractals, more-or-less-by-definition, should be scale invariant, but what's to say their Poser implementation is actually so.

    Thanks for the reminder about the 3D to 2D node conversion scheme.

  • @ghostship Possibly a good option, but not in PP2014... ;o)

    P.S. Although I have Poser 11 I still use PP2014 most of the time.

  • I had a sudden idea that rotating the object so that it wasn't aligned with the world axes might help...


    Bang goes that thought then !

    If anything it looks worse to me. Far worse.

  • @3dcheapskate Just thinking, if the cellular node is acceptably random within the 10 x 10 cells, can you try the fractal thing and apply an additional layer or two of scaled up cellular nodes with their outputs summed. I.e. one dark cell within the 10x scale will darken all the cells within one block of 1x scale, and so on. As I found with the same scheme of repeating tiled seamless textures as various scales, if you're at the right distance, it just looks random. My problem was that I was trying for that effect over a whole field, and the repetitions just turned into moirĂ© patterns at larger distances.

  • @anomalaus You may have something there. Using a second Cellular (10x the scale, no jitter) to drive the Jitter of the first cellular seems to give some improvement (N.B. I also increased the Jitter value on the 1st Cellular- that alone helps)

    Here's the extra nodes that I added, along with the existing Cellular.

    Note: I also increased the Jitter to 0.75(was 0.5 I think?) on the first Cellular, so if my maths is correct the actual value of jitter used will be 0.75 to 1.5, with the majority probably 1.0 to 1.25 at a guess.

    (In fact simply using higher jitter values with just the single Cellular makes the repeating pattern less obvious in many cases)

  • I think for this particular shader the repeating pattern is made most obvious when there's a small group of darker cells standing out amongst the more yellow cells. It's probably worth doing something tocut down on those darker cells.

  • Tried the 2 cellulars with a couple of slight changes...
    ...on a different prop ( http://www.sharecg.com/v/83581/gallery/11/Poser/Release-2-Poser-Stone-Shader-And-Room-Prop )...


    Still seeing the repeating patterns, but they're not so obvious - looking much better !

  • @3dcheapskate Yes, I think this looks better as well. This pattern doesn't appear as repetitive as the first one you posted.

  • Maybe we need a Penrose tiling (no pattern repetitions on an infinite plane) node? Anyone got an algorithm? ;-)

  • @anomalaus I was actually watching a video about that on YouTube a few months ago - the man himself doing a lecture at the RI.

    "Roger Penrose - Forbidden crystal symmetry in mathematics and architecture"

    (almost an hour long)

    Really interesting stuff, and extremely well presented. Old-school style.

    "...a drawing... or whatever they do these days ... with computers I suppose..."