Home Adobe Substance

Does anyone knows how extract curves node work in substance designer?

gnoop
polycounter
Offline / Send Message
gnoop polycounter
I want to make curve from an edge loop. The node  description mentions it can be done in Max Maya and Modo setting special tag  but I can find any nfo how.     

Is there any other way to import curves?

Replies

  • poopipe
    Options
    Offline / Send Message
    poopipe grand marshal polycounter
    Only have Max here but  I've managed to pass edges through

    like it says in the docs ..
    you need to make a selection set  - in this case from subobjects (edges)
    then when exporting the fbx make sure you have the selection set option ticked. 


    in designer add the object to your graph,
    enable As Subd button
    enter the selection set name into the filter box

    Here's the first problem.  The selection set name may not be what you entered

    I exported my fbx as ascii and had a look inside for the selectionset nodes.. 
    Collection: 1294050219920, "SelectionSet::Cylinder001_MySelection_Edge", "SelectionSet" 

    Max's docs explain that subobject selection set names will always be prefixed with the name of the parent object
    What they don't explain is that you also get a suffix naming the subobject type

    in my case the correct name to put in the filter box is this.. 
    "Cylinder001_MySelection_Edge"


    There's a big caveat attached to this.  In the form of the second problem

    when you subdvide your mesh in the scene node the edge indices get messed up and your selection becomes bollocks. I'm assuming this is a bug and have submitted a bug report
  • gnoop
    Options
    Offline / Send Message
    gnoop polycounter
    Thank you poopipe

    I recreated  your way step by step   and still see yellow triangle and no curve in SD  :# .    

    As always  Substance Designer does  it most crazy, cryptic, monstrously inconvenient way ever possible.    No hints whats wrong ,  zero  explanations .
    But I am not  surprised at all ,   their nodes are generally same  .  You understand nothing  until you re-made your own from scratch.   Had to make my own atlas scatter  just recently  because theirs is unusable.

    I especially  "love"  their pixel processor.  Zero preview what are you doing or what you pass through nodes.  Cryptic nodes names.   To use simple 5 grades   trigonometry  you have to break your head.    To simply  split or combine RGBA you need several  swizzles, vector floats.     Thanks god there is Blender where you can prototype before doing it in pixel processor .

      ps. For god sake why they wouldn't just allow to import an edge loop directly  as an object , or polygonal stripe . Isn't it obvious that current crazy way is a disaster.  

     
  • poopipe
    Options
    Offline / Send Message
    poopipe grand marshal polycounter
    The modelgraph stuff is all fixable, they need feedback at this stage or things aren't going to get improved though.

    It's definitely not production ready - I wish they'd marked it as experimental in the same way UE features are tagged, at least that way there's an excuse for the shonky documentation 

    as to the rest of your points . 
    honestly you just need to spend more time with it.  I do all those things daily and have no problem debugging now that I've worked out how to do it effectively.  Fx-Maps on the other hand are complete dicks to debug


  • gnoop
    Options
    Offline / Send Message
    gnoop polycounter
    poopipe said:

    honestly you just need to spend more time with it. 

    Yeah, maybe.   I very much prefer Blender nodes system vs  SD one.     Even  considering lots of  things are impossible in Blender.  

    1,   you don't have to convert or clean  grayscale to color , Blender does its guess on its own .     if you connect  rgb to vector input it  instantly interprets it as 3 vectors.  You could mix color  ,  grayscale and vector math nodes whenever you want.

    2 all the math is one single node where you could switch math operations   instead of re-connect another nodes

    3  you can use  shift+drag   to  make a dot node across several line  and  ctrl+drag  across to erase connections .   in SD  you have to first select a line  then push backspace .  And sometimes in a mess of lines it's hard to so.

    4 in Blender nodes auto-connect  when you put them over a line.

    5,  Blender math nodes   would  always  show  you something  with  an output connected . For vectors it would be colors  , for  scalar value something gray  you could remap to 0-1 to understand whats happening.      
    While in  SD you can also  make basic arithmetic   with regular nodes  in 32bit mode  and  see   how it goes  in preview , node to node  with normalized "auto level"   it doesn't work in pixel processor  where you often have no idea  why thing is not working as you wanted.     It's beside the fact that  switching to orange output  works slow  and  inconvenient  while in Blender you can  put gazillion output nodes next to each other  and switching in between them in a click .  

    6, I love Blenders simple  node grouping  and  value  sharing in between different nodes   while  in SD  you have to  build   super annoying   green  inputs  and "exposing" .      A month later you stop to understand  anything in your own  constructions.  

    In general  Blender node system is so much more  forgiving  to your minor mistakes    vs    SD one.     I always thought SD made  as much artist unfriendly node system as they possibly could.

     It became a bit better with time ( after they edded few conveniences  like dot node and Ctrl+E)  but still same in its core.

    I many times asked  for  quick output in the pixel processor  when you click on any node  and see output values . numbers,  normalized grayscale image or colors.    But I am sure this will never happen in my life time.











  • gnoop
    Options
    Offline / Send Message
    gnoop polycounter
    Another feature of Blender  node system  :    if you have a lack of available free space in between your nodes you can select all of them and hit S+drag    to scale them out of each other.
      It's convenient beyond words  and  in  SDesigner  you constantly  have to move them tediously  to blend in a new branch  or put the new branch  somewhere far outside  which turns you nodes into a mess gradually

     
      I wrote in Alegorithmic forum for decades asking them to take a look on Blender node system .   Thanks god they borrowed bypass command finally   , decade later  after a long lasting discussion  about how they believe it's impossible ,    and dot node at least.      There are yet so many to borrow.

Sign In or Register to comment.