Author Topic: Single directional triplanar mtl for multiple objects  (Read 2408 times)

2019-10-26, 13:57:42

mantaskava

  • Active Users
  • **
  • Posts: 80
    • View Profile
Hello,

So my goal is this:

Which is two objects with the same triplanar mtl (set on local) applied, however you can notice pattern goes vertically on vertical part, and horizontally on the horizontal (exactly how I want it to go).

BUT, before doing a whole series of actions it looked like this:


And now to get to my goal the only way I found out (so far) is these series of actions (6 of them to be exact):

Add Xform modifier


Select gizmo


Rotate gizmo


Rotate part back to it's place


Activate affect pivot only


Roate back(fix) pivot rotation


And so now finally we have what we want. But this workflow is far from being efficient IMO.
I tried uvw box mapping as well, but that too depends on local axis so it's basically the same problem. UNLESS you break the link(instance) between the two uvw maps (which I don't want to do) and rotate them separately.

Anyone has any suggestions on this?

Thanks

EDIT:

After doing these tests it looks to me that object's "local" axis and "pivot point axis (or direction)" is two totally different things, or..? Because if it were the same, rotating object's pivot point should change object's local axis too, which in result should change the direction of triplanar mtl. But in this case rotating pivot point does nothing.
So I guess we should first start from clarifying what exactly IS local axis and how does it correlate (or does not) with pivot point axis.
« Last Edit: 2019-10-26, 14:16:53 by mantaskava »

2019-10-26, 14:22:42
Reply #1

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8850
  • Let's move this topic, shall we?
    • View Profile
    • My Models
I can only tell how to shorten your action list, but it's not ideal solution either.

1. rotate horizontal beam vertically
2. reset its transforms
3. rotate it back to horizontal position

I don't know if there's better solution. Karba (FStorm) would probably add something like "align to longest axis" checkbox to a bitmap loader, but that wouldn't work with cubical objects :]
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2019-10-27, 08:39:22
Reply #2

mantaskava

  • Active Users
  • **
  • Posts: 80
    • View Profile
Yeah I already heard something about that "align to longest axis" feature in fstorm, and it sounds awesome. And it would work like 99% of the time, since how many perfectly cubical objects do we have in arch viz (or even real life)? :)
In Corona now it's a mess- you have to either create multiple copies of the same mtl just with different rotation settings in triplanar settings, or rotate object's transforms the way I (or you) described.

Quote
EDIT:

After doing these tests it looks to me that object's "local" axis and "pivot point axis (or direction)" is two totally different things, or..? Because if it were the same, rotating object's pivot point should change object's local axis too, which in result should change the direction of triplanar mtl. But in this case rotating pivot point does nothing.
So I guess we should first start from clarifying what exactly IS local axis and how does it correlate (or does not) with pivot point axis.

And do you have any comments on this? My main question still remains on what EXACTLY is local (axis/space?) ? If it's not the same as pivot point axis. And after these tests it's obvious that it isn't the same thing. Is there any way to "view" what local axis looks like? What's the direction of it?

2019-10-27, 10:41:36
Reply #3

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8850
  • Let's move this topic, shall we?
    • View Profile
    • My Models
Well, discrepancy between object's pivot point and its local transforms, is a mystery for me too. I guess that question should be asked in Autodesk community, maybe someone there knows the answer?

Yeah I already heard something about that "align to longest axis" feature in fstorm, and it sounds awesome.

Wait, what? Is this already implemented in Fstorm? o_O
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2019-10-27, 11:31:50
Reply #4

mantaskava

  • Active Users
  • **
  • Posts: 80
    • View Profile
I believe I've read a comment in Fstorm's group (dunno if it was from Andrey himfself or someone else) about implementing (or planning to implement) this feature into Fstorm.
BUT might be as well it was a dream of mine, not 100% sure now :))

2019-10-28, 17:27:24
Reply #5

mferster

  • Active Users
  • **
  • Posts: 523
    • View Profile
Well you could do a few things....

1. Create a reference cube with the correct mapping you want, convert it to editable poly > attach the other object that has the incorrect mapping, it should now assume the correct mapping then just delete the cube.

or

2.  if you use autoGrid  when creating your objects, it will assume the local cooridnates based on whatever surface you are generating it on.

or

3. Rotate your vertices in vertex sub object mode, then rotate it back in  position object selection mode. You could create a script to make this instantaneous.

2019-10-28, 18:01:13
Reply #6

mantaskava

  • Active Users
  • **
  • Posts: 80
    • View Profile
Mferster, and to add to your list:

4. My method that I described in the first post
5. Create multiple materials
6. Create all objects on the same plane/axis (for example front view)

So 6 in total (and I guess there's more) but none of them seems to be hassle-free. + Pretty much all of them has some kind of disadvantages.
But it looks like there's no other/magical way.

2019-10-28, 19:54:05
Reply #7

Juraj

  • Active Users
  • **
  • Posts: 4761
    • View Profile
    • studio website
Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2019-10-28, 20:42:49
Reply #8

mantaskava

  • Active Users
  • **
  • Posts: 80
    • View Profile
This feature definitely sounds interesting, too bad objects has to be unwrapped for this to work (as I understand).