Anton made me think with another post that he did earlyer about experimental projects in Poser like if its possible to make a vertex weight posable figure without haveing to re-write poser, so I started thinking how could be done, if possible . . .

In the image the magnets act like a lattice system to deform the geometry like a standard bone from poser would do, using ERC to control all the bones from the main body of the "figure" . . .

I'm gonna attach the file that I'm working on for u to play, please post comments and ideas, as this is just starting my guess is, yes u will need to have one magnet per vertex in the worst case, I'm trying to find a way to minimize the effort in doing that . . .

BTW, I discovered a way in which u can make a fully functional arm IK for AM just changing 3 values(well maybe more, say around 5) inside the .cr2 file, and AM will still be compatible with everything out there, I already sent the file to Anton . . .

Anton, are u around? Mind if I post the info here?

Apollo doesn't have IK in his arms?

Mmm . . . Nope, but that can be easyly fixed now, it's just a matter of hierarchy . . .

looks interesting....

inkyParent and nonInkyParent in the body part
and an inkyChain section in the Figure. :)

Message edited on: 07/07/2005 19:21

Hmmmm.... Is it possible two write a script to put a magnet on every polygon in a named group?

But what would be the performance cost of 40,000 or 80,000 magnets!? Hmmmm...

Dun worry for the magnets, since inside poser they are props that exist in the runtime folder, once all the work is done one can edit the .cr2 file to reference the components of the magnets to lower resolution props or even null props, u will only need the ERC control to dial . . . BTW, kuroyume, have u tried to set up an IK for AM in the arms or u r just saying how to create one but haven't done the attempt? 'cuz everytime someone tryies to set up a chain like: Slhd-UpArm-ForeArm-Hand u end up with weird results . . .

No I haven't. This is the standard practice for setting up IK chains in Poser. There may be some JP tweaking involved, but I suspect this has a lot to do with the initial orientation of AM's arms. Poser's IK solver likes the elbows slightly bent and the arms in T-form. You may need to actually break the more natural arms at 45Deg. on the mesh (and then in Setup) or more simply just do the Memorized pose with the arms out T. Maybe I'll give it a run. Regarding the magnets, there is a lot more info required for magnets than you reveal. Magnets are composed of three parts with all of their channels (Rotate, Translate, Scale, nOffsetA, nOffsetB, etc.). These take up buku memory space when talking about 40,000. Polygons only require an index and a set of vertex indices (maybe 32 or so bytes). Each magnet may require 1024KB or more (you're talking 40MB+ just for these magnets).

Cool, Only problem with "magnetic" joints is that spawning morphs will cause havoc. ANd there are some things about erc you have to keep in mind. 1)Poser can only handle so much of it before it gets confused. 2)It is possible to erase erc by saving figures back to the library in a certain sequence. 3)Erc uses memory. I would try to use a few magnets as possible. Is there a reason you need so many for the same effect? On side note: I purposely excluding arm chains.

I purposely created arm chains - not that I'd use them. :) They seem to work, though moving the IK chain via the hands seems to have too much hand reaction. Moving by the ForeArm works well. Before doing this (to a copy of course), I set the arms up, Memorized, and saved the figure as a new one (but rereferenced the original geometry obj). These are the new rotations: L/R Collar.UpDown: +/-10 Shldr.Bend: +/-15 UpArm.Bend: +/-15 And the Figure IK chains: inkyChain LeftArm { on name LeftArm addLink lCollar:1 addLink lPectoral:1 addLink lShldr:1 addLink lUpArm:1 addLink lBicep:1 addLink lForeArm:1 goal lHand:1 linkWeight 0 1 linkWeight 1 0.2 linkWeight 2 0.04 linkWeight 3 0.008 linkWeight 4 0.0016 linkWeight 5 0.00032 } inkyChain RightArm { on name RightArm addLink rCollar:1 addLink rPectoral:1 addLink rShldr:1 addLink rUpArm:1 addLink rBicep:1 addLink rForeArm:1 goal rHand:1 linkWeight 0 1 linkWeight 1 0.2 linkWeight 2 0.04 linkWeight 3 0.008 linkWeight 4 0.0016 linkWeight 5 0.00032 } Seems, as Anton did, you can bypass the naming of inkyParent/nonInkyParent. It may or may not make a difference to how the IK reacts.

Does your scaling still behave properly. I have seen IK interfere with scaling in some instances?

Do you mean figure scaling (propagatingScale) or scaling of the IK chain body parts?

ik chain body parts.

Seems to behave properly.

Why exclude IK in the arms?

I don't use IK often, but when I do, it's usually in the arms. Ever since I saw that Dr. Geep tutorial on posing with IK. :-)

Actually I just used 3 magnets to get the effect, I arraganged'em in a way they move the figure by slices, like a lattice in a standard 3d modeling program, now I have a new idea on how this could work without haveing to use one magnet per vertex . . .

Randym, Arm Ik has internal issues I wasn't willing to accept.