Rigid Follow Nodes - rotation craziness

clivewil2

hello.

i have used rigid follow nodes as attachment points for buttons or bows or buckles or whatever.

some of these rotate crazily when the part of the figure mesh is moved or twisted, often far far more than you'd expect from even a tiny figure deformation.

e.g. a common button placement point is somewhere between female breasts, and this one in particular can rotate up to 180 degrees at times. a major cause of this is re-sizing the breasts; i don't know about anyone else but 99% of the women i know do not have himalayan breasts, so i normally reduce them to about 50 - 60% of vanilla G8F size using the 'breasts gone" body morph (usually set between 35% - 50%) which causes the button to flip around.

i noticed it happens more chaotically if the clothing has a smooth modifier attached to it; so in some cases i reduced it by creating an invisible un-smoothed strip of polys from the original clothing that sat just behind it, and put my RFNs on that instead. a lot better but still not good.

the last thing i tried was creating a second set of RFNs on the back of the clothing and using a Look At constraint to point the front RFNs at each of those in turn - which is much more stable but still not perfect. (i might try combining the two workarounds - e.g. put the Look At targets on an invisible strip, but positioned only a couple of inches behind the chest instead of on the figure's back. wish me luck.)

so! after all this messing around trying to get them to behave, how *should* i have done them, is there a faultless method or do they always spin around?

is there anything i can to to the selected polys to make the RFN behave better?

what is better, a small narrow selection of polys (or even just a single poly) or a wide area of many polys per RFN?

Mada

Did you parent the RFN to the correct bodypart in your outfit? If its parented incorrectly it will often spin around like that, especially when you're posing the figure. Especially on the chest/upper abdomen there's a narrow range on where they will work best and it takes a bit of experimenting to find the right spot.

There is no set rule for how many polys to use, its a case of testing to see what works best. I've had good results with a single polygon for some outfits,  but a wider selection worked better in other outfits.

clivewil2

Mada said:

Did you parent the RFN to the correct bodypart in your outfit?

i went purely by the position the button needed to be in - once it was placed in the correct spot, whatever geometry was underneath it became the parent for its RFN - typically Chest Lower for the most common and troublesome example, or otherwise Chest Upper depending on the clothing article in question

clivewil2

ok i was a victim of my own assumption, i didn't look closely at the transform gizmo - it seems that one RFN will not point at another **

the button itself will though, or a poly/null with the button attached to that, and it tracks beautifully - full range of figure movement with no weird effects.

i'll try and simplify it so it isn't such a mess, but at least now i do have a viable workaround. 

**  (i am using a second row of 'rear' RFNs as Point At targets for the front ones)

clivewil2

i just wanted to post an update to this issue, i seem to have found a solution - i just thought of it and tested it, and haven't managed to severely break it yet.

instead of selecting a contiguous area of polys, i selected four individual corner points instead, wondering if the node's rotation would be averaged out between four separate locations rather than a single area being averaged out.

it seems that it is the case, and in spite of various figure deformations and extreme poses, the button on the 4-pointed node stayed pretty much level even though the other two buttons - one attached to a single poly and the other to a 'quad' of connected polys - have both gone haywire, as you can see in the pic (the figure has Breasts Together, Hip Bend Back, Torso Bend and Torso Twist applied simultaneously.)

pic shows the polys i used to create the new node.

apologies if this idea was already common knowledge - i never saw it written up elsewhere, but then again i didn't really look that hard.

 

---> note - the node is applied directly to the shirt itself, no need for any other supporting geometry that i mentioned in previous posts, AND the shirt has a smooth modifier on it - which was usually disastrous with the single area nodes.