It's really strange.   It tends to work better in a bit older versions of Poser.  I just tested it in Poser 7, and it works fine, so like you are saying, it's probably fairly new. I recall it being problematic in P8/2010 as well.

Ok I've emailed it to one person, just drop me a sitemail with your email and I'll send it.

Here is a description of how it's set up:

Two obj files, one with just the base hair, and one with a soft decorated group. It's quicker to just test with the base hair first.  It has the following groups:

"constraint" for the hair roots to keep them from falling
SC1_auv for the skullcap, probably works best constrained too

"Hairbase" for the main hair, and "Hairsoft1" for the full hair, a test with soft decorated groups to add thickness to the hair.

The skullcap is made in Wings, and the strips were made in MD2 and then attached in wings. I found that the hex triangles that I get from MD are -way- faster than quads.  The full hair with soft decorated groups is 20k polys, and gives no problems as far as simulation speed. Could probably double it and it would still do fine (just might, if it ends up working, with more thickness)

The main problem that arises is if you run a simulation where V4 tilts her head or moves to the side, with some sort of invisible bump there. It shows clearly in the image in the first post.

Also, please feel free to use the skullcap and hair strips to make your own hairs!  All I ask is that the resulting hair is made freely available for the community.  If you want more hair strips I can make them easily.

I've tried SR3 and it made no difference. The collision offset hasn't done anything either.  What's the best way to send you the obj files, want to PM me your mail adresses, or is there another method here that works better?

Ok, revisiting this thread with some updates, and a challenge!

Earlier in June, I spent over a week with SmithMicro, who did all they could to help me test and resolve the issue, but we weren't able to do it. It seems to be an issue specifically with the Generation 4 figures (happens to M4 too), which is what I want to make the hair for.

I've got really nice ideas for this hair, got photo sources made and everything is set, but this issue still persists. One option is to make each "piece" thinner, that way they would probably just slide around the bump, but it wouldn't be a great solution, especially for animation. 

If I can finish the hair, it will be available completely free for the community. I'm making it because I'd like nice, flowing -long- hair that doesn't take 3 hours to pose with handles, or with the really long calculation times in the cloth room.  I am sure there are others who would enjoy it too, if it turns out well..  So is anyone out there interested/willing to help out testing or troubleshooting, figuring out a workaround?  If you are, just lemme know and I will send some test props! It'd be cool to see if we could make some really good dynamic cloth hairs for the Poser community.

Yep, aware of that. :)  This howerver is a special refit of pants that I have made for V4, that I am refitting to M4 in another 3d program, while preserving the UV coordinates.  I want to present them in normal 100% scale as the props are loaded, with constrained groups intact. This has worked fine, I just needed a formula to get the scaling just right.

Quote - Actually, there is no need to scale the dynamic clothing.If a dress fits a figure which is scaled at 100% and saved as a smart prop,then if it is applied to a figure which has been scaled down, it will automatically rescale to fit.

Alright thanks for the tips, should get what I need!

Quote - try deleting V4's eyebrow bone

Yeah I tried all sorts of mutilation, but the bug persists even if I disable the head in collisions and only collide it with say the chest.

Choreographed didn't do it either. The constraints still won't follow the parented skullcap. :/ It also raises another problem, which is the rest of the body. It's a very long hair, so it needs to drape over arms, chest, back etc. If any V4 bodypart is selected, not just the head, the same weird thing happens.

SM is on the case at the moment trying to help figure out what's going on and finding a workaround, too.

Quote - Constrained groups do not works against props. You can make it choreographed.

Maybe it is a bug, but if there is any magnet different Poser version can select different way child props. Default setting should not include magnets o colliding group, but maybe P9 does not recognize it as a magnet.

Nevertheless I think that transparent prop resolve problem faster and easier then SM fix Poser. Moreover you can join hair strands and head phantom as a 1 prop (choreographed group)

sample

Ok something really weird happened. I was messing around with turning on and off different collision objects, and suddenly it started working. When I reloaded the project again and did the same type of things, it was back to it's normal weirdness though. Can't for my life figure out what made it suddenly work.

Quote - I think that there can be any JMC magnet. Try craeate phantom head - head shape prop but less resolution. It really helps. V4 head has number of polygons on ears, lashes etc. Their normals have different directions and it can cause simulation errors.

I made a test with this, using the skullcap parented to the head for the simulation. Turns out that the hair's constrained group will not follow the skullcap unless the head is included in the simulation too.

It turns out that JoeBushido's cloth hair doesn't have this issue. It's a conforming item with the hair included as a prop which is then simulated. Why this works is beyond me though.

If I want to continue working with the hair, I'll have to stick to Poser 7.  Saving a morph won't really do much since it's a cloth hair.  Reopening the project again in PP2012 to work on materials and so on would work though.

The idea has been to make something that can eventually be distributed, but it'll be hard to make anything that can be distributed if people are forced to use a certain older version of Poser.

I've started a ticket with Smithmicro, providing them with what we have so far here.

Tried that.  Seems like there is a -piece- of the head that it collides with, not the whole thing. I made another experiment and made the whole hairpiece constrained. Even then, the portion where the dynamic group would collide with whatever it is in empty space did the same thing to the constrained group, like a stretched out block about the height of her ears...