PDA

View Full Version : Will vanilla joint rigs ever come back?



ArwingXL
06-25-2012, 02:46 PM
Now, i don't want to start a flame war. There's plenty of you who'll jump up on their boxes with a "THERE'S NOTHING WRONG WITH THE IK." I know you guys already. you don't use joint goals. that's fine. This isn't about you. Your methods work fine, your 3rd party autoriggers work fine, ik booster works fine, and your 3rd party plugins work fine too.

This isn't about those things. This is about Newtek's vanilla ik solver after 9.3.

TO THIS DAY:
Whenever i work for a new contact with lightwave, i have to send them mdds. Want to know why? New clients want to use Lightwave 10 or 11, and you know what I do my ik rigging and animation in? Lightwave 9.3.1. because it was the last version after 8.5 where you could attach a goal to a joint with "keep goal within reach" and have an intuitive way to control elbows, spines, knees, tails and what not. End goal with a strength of 100 and an elbow goal with a strength of 1. works really well, and nobody has complained at all about pops or issues. End of story. If you take this exact same type of rig, pure vanilla lightwave mind you, and import/make it in 9.6.1, 10.1, or even 11, it becomes almost useless. I say almost, because there's been 1 or 2 occasions, where an elbow or a knee goal has worked properly when the bone resting positions were at some magical bend.

I don't mind doing all CA in one version and rendering in others, but deep down, I just wish Newtek would unbreak the joint goals so I could just stick with one version completely.

RebelHill
06-25-2012, 03:43 PM
The alteration you're referring to was a rewrite of the IK solver. It most certainly IS more robust now than it was previously, and solved many of the pop/flip issues of the previous solver.

I very much doubt its gonna return, so if I were u Id just leave it to history and get familiar with the new stuff... IK itself aside, there' just SO MUCH more in recent versions for motion control/constraints/etc that Im sure ull find more stuff to be glad of having than stuff that u miss.

stiff paper
06-25-2012, 04:14 PM
I do sympathize, because I personally despise nodes, but there's a vociferous coterie that seemingly would be perfectly happy if, when you ran LW, it opened up as a node graph and nothing else.

(Do not bother. I am deliberately poking the hornets' nest with a stick...)

The IK really isn't going back to the way it used to be though, because the way it used to be was terrible in a lot of circumstances.

I think the answer is that you can't use the new IK solver in the same way you could the old one, and it'll lead to problems if you try to. Learning the "new" method isn't really very difficult though, and once you've got the hang of it, it isn't any slower to set up or any more difficult to use.

Edit: RebelHill has links in his post to his RHiggit! site, where there are some free rigged models, and then on his YouTube channel there are some vids showing how to use the rigs. You might not need anything as fully functioned as an RHiggit! rig, but the examples might help to convince you that the new way of doing things works at least as well as it used to... it's just not exactly the same.

SplineGod
06-25-2012, 04:52 PM
I have to agree. The newer IK solver seems far more stable then the old one. The newer constaints are great and actually work. There were times when I would almost tear my hair out with trying to use some of the older contraints which always seemed to have gotchas everywhere. To me the newer IK and contraints are far easier to use, are in a more logical place, and most importantly - they work :)
Do you have an example you might be able to post showing the difference? Maybe video showing 9.3 compared to 11?

ArwingXL
06-25-2012, 05:58 PM
I can understand why the rewrite for the solver was done, and am familiar with working with the new system's constraints and tools. If I look at a project and it only needs a few seconds or a minute or two of animation, I'll stick with 9.6/10/11's system, and work with single goals, and constraints/nulls/handles/etc. I work from pretimed frame sheets like the ones 2d animaters use, so it just means taking a little longer to cross off keyframes I've covered, but if its something more like what I'm currently working on which is 30 minutes with a max of around 5 characters per scene, less controls usually make the experience happier.

Rebel Hill's rigs look like they're made for grab n pull/rotate, so i may teach myself how to key them sometime in the future.

erikals
06-25-2012, 06:23 PM
 
you really should take a look at his RH rig and rig tutorials (check his signature)

another thing i like about RHiggit is the way the animation controls are set up, fast, efficient and user-friendly.
(or at least to my experience, no CA pro just yet...)

 

Dodgy
06-26-2012, 01:46 AM
I actually agree. I did the exact same thing for elbows and knees and it worked fine. Goal Strength really doesn't seem to work anywhere near as well as it used to, a GS of 100 on the hand would keep it pinned to the hand goal perfectly and still point the elbow at the other goal. Now I have to boost that to 10,000 and the hand still slips a little, but it depends on the joint layout. All the rigs I developed pre 9.3 just stopped working after that and I had to replace them all as I came to use them. The new IK system has a lot of good stuff in it, but at the time it was very annoying, but I've got used to it.

RebelHill
06-26-2012, 04:11 AM
The whole elbow knee goal thing was really a way to get around LWs sometimes flippy IK I found, but still not always perfect (there were still issues of gimbal lock at times on the first IK bone). But the new system, being designed much more along the "planed 2d IK" approach is not only more stable overall, but also just as easy to control for getting elbows etc correct.

Either rotate the IK plane, and there's a character space limb twist, or pole it, and there's a world space point. Or do as I do, and make it blendable between the 2.

Im not entirely sure what you mean by "pull/rotate rigs" though... Im not familiar of another way of keying/animating goals, or other items.

safetyman
06-26-2012, 05:52 AM
Im not entirely sure what you mean by "pull/rotate rigs" though... Im not familiar of another way of keying/animating goals, or other items.

I have about as much knowledge of rigging and animating as a brain-dead ant, but I've been using Blender and it has a very simple system for moving/rotating goals and joints. You turn on the rotate button which shows the rotate widget, but you can also move the goal without having to change widgets, so moving then rotating is a snap. Maybe this is what he's referring to? Forgive my ignorance.

ArwingXL
06-28-2012, 01:33 AM
safetyman has the same idea as I do. This rig is the same type I've been describing. It was drawn up on paper first, then built to be very clean, speedy, and minimalist as to function as one would expect a stop action puppet to function so i can focus primarily on keys, extremes, and breakdown frames and not the process of making those frames.

That said, I've already fixed the problem after running 9.3 and 10 side by side with the same rig and animation running on loop.

This is one of the rigs affected by the issue. While it fell apart at first in 10 and 9.6, it also began working almost exactly the same when i switched the elbow/knee/tail/neck goals to softIK and set the end goals strengths to 500,000. The "almost exactly the same" kicks in mainly due to goal behaviors in the new ik rewrite. A joint goal held by the joint itself no longer keeps its juxtaposition in the ik chain as a character moves through an environment without either being parented properly to the character's skeleton itself or being keyed along with the end goal to make sure joint orientation is consistent to the actions a character is doing.
http://radcore.us/DUMP/grabnpull_rig.jpg

RebelHill
06-28-2012, 03:27 AM
it also began working almost exactly the same when i switched the elbow/knee/tail/neck goals to softIK

Now thats interesting. Something to experiemnt further with there.

ArwingXL
06-28-2012, 12:22 PM
the weird thing is everything but the neck survived that conversion(but not the tail, which is odd). Ended up just replacing the neck ik with plg_curve_bone spline ik, which works a lot better. also, double hinged legs get interesting to work with on rigs like this, because the rig screams at you if you don't key all of the goals at once, if only to just grab one joint and wiggle it a little to autokey its location.

jeric_synergy
06-28-2012, 01:20 PM
off the point....


the
Dude, your avatar is incredibly distracting. :screwy: :cry:

ArwingXL
06-28-2012, 02:30 PM
fixed.

jeric_synergy
06-28-2012, 03:19 PM
fixed.
Seriously, THANK YOU. Trying to follow a high-level rigging discussion with the old avatar running on the side was killing me. :thumbsup: