Warning....rant follows.... Hi So I've been trying to get some realism into my renders, and I've got to the point where the skin textures are not bad, but realism is being let down by poses and joints. In particular the elbow, shoulder, knee and hip joints. So I'm thinking, hey, why not develop some better JCM's to fix the problem. So I investigate. No go. The problem with JCM's is that say the elbow only gets screwy once it's bent more than 100 degrees, and JCM's start applying from 0 degrees, so you can't really have a JCM that starts effecting things after say a joint rotation of 90 degrees. So then I think, what about magnets. So I do the old "pose V3 with her shoulders back and arms down and look at the poor joint". The inside of her arms go into her torso. So then I lay down some magnets to try and fix things. Now I might be wrong here, but it seems to me the magnets don't work too well when there is a JCM (the std one's with V3) being applied to the area. So I couldn't get anything resembling a good result. So....next step is to export the previous mesh as obj into Maya and start to try and manually fix the shoulder joint by playing with vertices. But....Maya is designed to work on low poly meshes for deformation. A 60k poly mesh becomes and unworkable mesh. There are some tools to help (blendshapes, clusters, lattices, etc), but they don't really apply to fixing a hi-poly mesh like V3. Totally frustraited, I start playing with shaders in Maya and come up with something pretty good for skin. Run the render - looks a treat. But there is one little problem, V3's bottom is not quite right. I render a few different angles and nothing quite fixes it - so I head back to P5 to morph it into shape.....and......P5 crashes. Agghhhh. So it seems that P5 has a bunch of limitations which people have devised sneaky workaround for. But there are some things (joints!) where P5's implementation means they will be very hard to work-around. I ended up posting the final Maya render, at http://www.renderosity.com/viewed.ez?galleryid=774523&Start=1&Sectionid=26&filter_genre_id=0&WhatsNew=Yes Thanks for listening - had to get that off my chest.

Days like these, you can't help wondering if it's worth all the trouble... Your picture came out well, if it's any consolation. It shows that the end is more important than the means. :) When you consider the origins of Poser, as a quick and easy drawing aid, it's not surprising that later versions are having to swim against an increasingly strong current. It was never designed as a realistic flesh simulator, and it amazes me that people get the results that they do.

I have a bunch of multiple magnet created, joint controlled morphs that have exactly the delayed onset you're looking for. The concept has been discussed in a few tech forums. There are some excellent tutorials and I should always mention the debt we all owe to Stephen Stahlberg... The hardest part when having one limb deform another (shoulder-chest, upper arm-forearm, thigh-shin) is how to spawn the morphs without including effects of any existing JCM already on the figure. In all the JCM cases I have seen, none have any deforming effect when the figure is in its zero pose. The simplest way around this is to duplicate the magnet parented to one body part onto the other body part sharing the effect. The caveat, of course, is to have the magnet base scales identical (which they won't be by default on different body parts being derived from the size of that part). That way the to magnets will apply the same amount of deformation to the adjoining parts. Once the magnets are successfully placed to achieve the desired deformation in the maximum joint bend position, the figure can be returned to its zero pose and the morphs spawned without interference from any existing JCM. To delay onset of morphs and scale their effects, you can use a dummy (no deltas) morph with a negative initial value (the offset) depending on the actual joint rotation. The other morphs have their limits set and depend on the dummy. As the joint bends, the morph limits prevent them taking effect until the dummy offset is accounted for, then the morphs commence their deformation. In the attached sample image, (Delenhoure GND-orion1167, NB Bikini-...) the morphs include: Buttock GluteBendFix - uses Buttock xrot ThighBulge Thigh KneeBendCtrl - the dummy control morph using Shin xrot ThighBulge KneeBulge KneeCrease CalfSquash Shin KneeBulge CalfSquash Each morph was created from up to three separate magnets per body part. I have other examples of extreme joint bends on R'otica in the Main Gallery as GeoffIX

I should probably also mention a couple of rules of thumb I have learnt, abandoned to try something different, and then relearnt along the way.

[Disengage egg-sucking tutorial mode] ;-)

I always zero and lock side-side and twist rotations of the forearm and shin. This avoids nasty broken knees and elbows but makes IK generated poses an imperial sceptre up the backside. The desired pose can still be achieved by rotations of the hand(+shoulder) and foot(+thigh), since their deformation zones only affect the wrist and ankle, not the whole forearm or shin.

V3's Buttock/Thigh bends take on strange appearances if limits are not adhered to . I find the least objectionable way to bring the knee to the chest is bending the thigh alone to its limit of 65, then adding buttock bend to its limit of 30 before resuming thigh bend the rest of the way.

Unfortunately, any buttock bend deforms the groin area of the hip, resulting in even more compression of the already limited mesh in this area. I have had some success compensating for this with magnets centred at the y,z buttock bend coordinates bringing the pubic region forward as the buttocks bend. It's by no means perfect, but third party genitals can retain their shape and recognisability ;-)

And the final rule of thumb...

If in doubt, throw magnets at the problem ;-) (I wish Santa would bring me a real modelling tool...)

If anyone's interested, I'll see about bundling up the morphs into injectors.

Prefix-GeoffIX-Suffix

Message edited on: 12/12/2004 04:16

And the biggest wish... That Curious Labs would find the wherewithall to extract their appendages from their orifices and rewrite Poser for the Mac to use MachO binaries rather than CFM so it can work with a version of Python that supports Tkinter, GRRRR!!!! NOT holding my breath.:-(

"If anyone's interested, I'll see about bundling up the morphs into injectors." I am definitely interested :-) Looking at solutions found by others is the best way for me to learn more about Poser.

Ahhhhh, Rederosity ate my original msg AGAIN!!!! Anyway, thanks a heap for this info gwhicks. It may contain the breakthru I need, although I've found a workaround by exporting the mesh to another 3d app and fixing the joint there. Time consuming, but very effective (see my latest 2 gallery images). Thanks again - great, detailed info.

I've taken to copying my post text to a text editor before I click any browser buttons so as not to lose what I've typed. It's seriously disconcerting to have to recreate a chain of thought or revisit a stream of consciousness to redo a swallowed message. Oh, and I abhor the fact that IK posing ignores the zeroed twist and side-side limits when turned off. GRRRR!! I'm NOT paranoid, I'm PERSECUTED !!! ;-(

I shall have to check back -- Geep reposted some tutorials on posing recently, and covered some of this unwanted-twisting stuff.

But I recall being told that the knee joint does have some side-to-side movement if it's bent.

Message edited on: 12/13/2004 10:09

AntoniaTiger, I have no doubt that is anatomically correct, but in the Poser realm bent knee shin side-side is better (less mesh break up at the knee) represented by thigh twist IMHO. For example, if the shin bend is 90, the side-side bend zones don't affect enough of the parent thigh to avoid strange mesh effects. Whereas twisting the thigh rotates the bent shin child without any distortion while distributing the twist along the length of the thigh smoothly. Anyway, (pp.) "They're more guidelines than rules" and YMMV, but I've seen enough extreme knee bend poses with large shin side-side components leaving the calf muscle in impossible positions to want to try for something better :-).