Okay, I've currently been making characters the hard way -- Making CR2s from either hacked existing CR2s or, call me crazy, from scratch. However, it's not easy to make one *totally* from scratch, and though I had bad experiences with the PHI method, I went ahead and tried it again. I decided to go for something infinitely simple -- a bendable stick (chamfered-cap cylinder in seven pieces). That and I wanted to examine the CR2 it made to determine how to put curve dials in. Now, I tried it with both a curve version and a non-curve version. The structure goes like so: SegR3, SegR2, SegR1, Seg0, SegL1, SegL2, SegL3. The non-curve version works fine. The curve version works fine *except* for the joint between Seg0 and SegR1. If I switch Curve to 0, it works like the non-curved one, and looks okay but not curved. If I leave Curve at 1, a weird thing happens: SegR1 seems to not actually curve, but to recede in the distance. This just sounded to me like a JP issue, but it doesn't seem to be because -- check this out -- it seems to recede *from both the front and back*. If I view it from above, it's fine, no receding, it's not bent. If I view it from the exact front or the exact back, SegR1 doesn't seem to be connected to SegR0. If I angle up a little, it's receding back, and SegR1 is *always* behind Seg0, from either side. I should add that the problem doesn't happen to SegL1, on the other side with the same settings. Has anyone else encountered this? Can you tell me how to get around it?

I did find that the 'break' goes away if I tweak the centre -- for some reason the PHI import doens't seem to have set the joint centre in the centre... But it still doesn't curve. In this image, the parts are symmetrical -- zRotate is set to -45 on SegR1 and to 45 on SegL1 and everything else is left alone. The sides *should* look the same by all I can find. The other JP settings do match and don't seem to be off the say the centre was.

If it makes any difference, here's the PHI file:

objFile :Runtime:Geometries:Test:7PieceStick.obj

1 seg0 xzy curve
        2 segL1 xzy curve
                3 segL2 xzy curve
                        4 segL3 xzy curve
        2 segR1 xzy curve
                3 segR2 xzy curve
                        4 segR3 xzy curve
ikChain RightSide segR1 segR2 segR3
ikChain LeftSide segL1 segL2 segL3

Okay, now this is completely, utterly weird... After posting that I quit Poser and was doing some Photoshop stuff for a while. Then i went back into Poser to show my wife the weirdness and it didn't do it. It seemed fine. I could twist it and shape it and it maintained the curve. Only the thumbnail was there to prove I wasn't crazy. So now I have the most useless prop I've ever created -- a fully posable pretzel.

BTW the hard seeming edges you see in the pretzel pic were simply because I used very few vertices. Only two between each seam.

lol!! umm... well, it isn't receding. it's stretching down to meet the end of the sr0 thing. anthony probably has a 'curved part bug' report for this sort of thing. or, basically, that curve stuff works weird in any case. it could also be because you don't have enough segments per each segment to curve and bend them properly. skimping on test figure geometry can lead to bending and posing problems.

Oh, I knew the curve wouldn't be smooth. It was just there to get the curve dials out of a controlled file that I knew what was there. But that doesn't explain the original weirdness... because the geometry was symmetrically identical on both sides and it only did it on one side. And the right side didn't curve at all. At first. Until I restarted Poser, apparently. As to the other thing -- If I believed I could make it do it again to make it worth the effort, I'd make an animation to show you scrolling around it. It was deiniftely showing it behind from both sides, and no effect from the top and bottom. I swear, this was the weirdest thing. You're looking down from a 3/4 above view in that first pic. From the side, they lined up perfectly. From above they lined up perfectly. As I approached any 3/4 angle, it did that. And rendered that way, too.

Curves often do wierd things. Here's a few rules I always follow when working with curves: 1) Never put a curve dial in the first or last segment in a chain (If you do then set it to zero, which is the same as not having one. After setting it to zero, you need to memorise the body part) 2) Never put a curve dial in a segment which is next to a hip-equivalent body part (seg0 in this case). 3) Never put a curve dial in a body part that doesn't have rotational symmetry. Sometimes you can break some of those rules without seeing any ill effects, but bad things can happen unexpectedly later on if the user parents the figure to another one, applies a pose or something like that.

But then how do I pull off the Boid (Boa, Python) figure I intend to make? The rattlesnake doesn't look anything like a boid, so I don't want to use that (besides, it has permanent rattles) and I have a Red Tail Boa and a Reticulated Python that, with a little soothing, I might be able to get to lay still long enough to be scaned to sample the pattern from (if not there's always the webcam). But, conceivably on a snake, the defaultPick should be the head or the segment 1 or 2 back from it. That rule would prevent assigning a curve to the segments on either side. And a snake body is close to radially symmetrical, but not exactly. Actually, doesn't the P4 horse also violate all of those rules?

When I did my snakes, I made two versions - one with the head as base and one with the centre segment as base. I followed the rules above on both. They look fine, though in some renders you can just see the absence of curve dials in the central three segments. Since there are thirty something segments overall though, it's not really that noticeable. The horse does break those rules, yes. I said they're the rules I follow, not the ones other people do. I often want to do odd things with my stuff, like parent a tool figure to the final segment of a tentacle and be able to apply a MAT pose to the tool. Under those circumstances a curve dial will make some really strange things happen when you apply the MAT pose. Probably nobody wants to do that to a horse tail. I'm also usually making things like tentacles or ropes etc, not horses, so my approach is different. You've already seen what happens when you put curve dials either side of the hip equivalent in a snake/tentacle/rope type of figure.

Yep, though the weirdest part was that it only happened at first, and after restarting poser it worked fine. I gave the posable stick away as part of the consolation on my Scarecrow contest that i just finished judging a few hours ago, describing it as a fully-posable pretzel. As far as the not-enough-vertices thing went -- that's the easiest thing to fix -- All I'd have to do is import the geometry, refine in a few places along each segment, and export over the top of the old OBJ and blow away the RSR. I was a little confused as to why the JP editor didn't have any dials avialable than the xrot (twist, cause I made it horizontally). My first thought was that the bend(z) or side-side(y) was bad but there was no option for those except on Segs 2 of each side. Well, If nothing else, I can now make nungchaku, seven-piece-rods, and tentacles from this experience B^)

Hmmmm.....not sure about the missing bend and side-side JPs. I can see why seg0 might not have them, since bend and side-side are just rotations that don't include any joint style deformation of the body part. Can't see why it would keep twist though. The others not having them is really strange.

heyas; my solution is not to use curves. the reason the jps are not showing up is because you are using curves. when you opt to use curves, you're letting poser handle those jps. or it's using a standard set of jps, or something bizarre. you CAN insert the jps into the cr2 and then be able to edit them, with the curve still in place. most people don't use curves, because they want to set their jps. ;)

Ahh, but remember I was using curves to know what they looked like in a controlled environment so I can add curves to figures I hack togehter in gvim! Since I now know what a curve dial looks like, I can set one up in another figure by hand and have absolute control of it.