Transforms as attach nodes

[SasquatchM]

If you are using a transform to define an attach node, which axis will the node connect along?

[Ledenko]

Does this even work yet? I tested it, tried to add a node along with regularly config-defined node and couldn't get them to show up.

[SasquatchM]

Well, it was supposed to work as of .20, but I haven't tried it yet

[Porkjet]

It works. I only got it to work after looking at Alskari's hollow structures and hulls mod http://forum.kerbalspaceprogram.com/threads/29031-0-21-x-Hollow-Structure-Hulls

This is how he set it up:

NODE

{

name = CONNECT2 // reference name of the node. Can be anything you want.

transform = TF2 // name of your unity transform

size = 1.0

method = FIXED_JOINT //FIXED_JOINT, HINGE_JOINT, LOCKED_JOINT, MERGED_PHYSICS or NO_PHYSICS

}

Your transforms BLUE axis has to point towards the attach direction.

[NathanKell]

The problem is that IIRC the size parameter doesn't work, and anyway if one does this it's no longer cfg-editable in position (can be a problem). What is the advantage, anyway?

[SasquatchM]

The main advantage I can see is that you can set up all of your attachment points ahead of time without having to figure out coordinates.

I am currently reworking my old sms plane parts pack and to define all of the node manually for 50 or so parts takes forever.

[NathanKell]

Oh, I see. Hmm, I'd just add a helper object in max, and note its position and direction; it's just the extra step of writing down the transform's position once made, rather than exporting that directly?

[Ledenko]

Just today I was messing with an attachment node that was supposed to be slightly rotated to fit a particular piece correctly... In the end, I put a number in config that I got with voodoo and it worked, but it would have been much easier to just plop down a properly oriented transform. So I can see uses for them.

[Greys]

what is this NODE{}? we were told to use ATTACH{}, is NODE{} a synonym or a completely separate alternate and maybe more functional version of the same thing?

I must have answers! I must have hot chocolate!

Anyways, there were a lot of little problems such as Nathan's 'node size is ignored'; Razchek and Sirkuit (I have no idea what the spelling of that is) went deep and discovered that joint time was being ignored back around 0.21 or so, and would always be FIXED_JOINT regardless.

I seem to recall as well that the joint's physics didn't operate quite the same, but joint physics are a mystery box of love and death anyways.

I have heard nothing about Squad even touching it since it's release in 0.20, though I'm not always the most giving-a-****-what-squad-did-on-wensday person so I may have missed something.

Another advantage of using transforms based attachment nodes is that sometimes it's very difficult or technically restricted to make a transform at the angle you want. I don't know if it still is, but there was a time that if you specified rotation along more than one axis for a node; KSP would round the rotation to a 90 degree mark, so 0, 0, 0,0.7, 0.7, 0 would give you 0, 0, 0, 0, 0, 1; so using transforms will properly allow you to have any aribtrary orientation you desire.

Furthermore matched with MODEL{scale}, if the nodes are defined by transforms, and those transforms are part of a model that is scaled by say scale=1,2,1, they will automatically be scaled by 1,2,1 too; otherwise you would need to redo the nodes manually for any irregular scaling factor that can't be treated as a regular scaling factor.