Works in Progress

#421

Hello, Joe.

I’m curious about this problem, and about any solution you might have in dealing with it. Last November I sent a bug report to Steve Sappingtion about the jittery behavior that can result when a two bone chain is nearly stretched out, & where the distal bone is governed by a Kinematic Constraint. Bob Croucher replied. Bob talked over my head, about “Support Limit Solvers” and about “fairly low precision numbers.”

Well, anyway, the upshot was that Bob devised a new way to calculate kinematic chains, and that this new way was implemented into v11.

I’m still using v8.5 and v10.5. I haven’t upgraded this year. I’m assuming you’re using the latest version of v11. [Of course you are! – you’re using the latest dynamics constraints! – how could I forget!] It’s vexing to hear that this sort of jitterbug problem is still around. Please, Joe, if you do find a solution or a workaround, could you share it with us?

Stephen Millingen was having similar troubles with Briar Rose. It’s something about beautiful women characters. They have to stand on their long lovely legs, keeping them fairly rigid, as they shift their weight around. That situation, where the Kinematic chain between the heel and the hip is almost fully stretched out, is very taxing for the program to calculate. Cartoony characters, such as my alligator boy, can always get away with keeping their knees slightly bent. When the knees are bent the jitterbug problem vanishes. Beautiful babes don’t have this luxury.

Have you got a minute, Pequod? Care to chime in? How is Briar behaving these days? Legs doing okay?

Sincerely,

Carl Raillard

#422

Well, I was having problems in 10.5 with some jittering on some arms. This problem is now resolved in v11. Not wanting to speak for him, but I think that joe is having problems of a different nature… not sure what though.

#423

Hi Carl,

Hmmm - weird. Well, that is pretty much the problem I’ve been having - I probably spent 70% of my time tuning out the little “jerks” and I had to settle for a bit less leg extension than I wanted. It worked out OK because of the design of the character - the walk looks a touch like a “march” - a very purposeful walk, anyway - but if I want to do a more languid, sexy walk, I may be in trouble. Often the amount of difference between having a pop and not having a pop was very, very small. I did a lot of zooming in, setting markers, advancing keys and using the arrow keys - very tedious. I think it may have something to do with the fact that I set my bones up with a “pre-bend”… but I’m not sure.

If I do figure out a way around it, I will post it. I may also get hold of Bob and see what he has to say about this little problem - or if he knows of a way to work around it.

I think that a rotational limit constraint might work to an extent - but in AM I’ve found that they can introduce their own problems when combined with other constraints…it would be nice to get this figured out - it would effectively make me 100% faster as an animator…

I’d like to know if Stephen is running into the same thing with Briar Rose…

JoeW

#424

Carl, Briar’s legs are finally behaving themselves, the new ‘IK solver’ Bob introduced is one of the best improvements to animation in v11.
I could be wrong, but I think Joe, you are having a slightly different problem. With the pre-version 11 releases, Briar’s legs or arms would do a fairly decent impression of Elvis if they were fully extended and crucially had long gaps between keyframes. Basically, if her limbs were motionless for extended periods of time, the tremors would start.

Joe, great walk cycle. Personally, I hate doing semi-realistic walk/run cycles. You tend to lose sight of the animation and just concentrate on ironing out glitches. I would however, try to get that leg extension back in. I’ve noticed from past observation that although the snap that happens just after the straightened leg hits the ground as the knees start to bend, can look a bit severe, (and you do have to spend ages softening it) it does help to convey the feeling of impact and weight. I think you need at least 2 frames where the leg is straight.

The only animation I could find of Briar done in v11, where you can see her legs is this: (BTW it is in no way here to show you how to do walk cycles!)

www.hash.com/users/milos/walkintomill2.mov

The scene is actually a lot darker normally, helps hide the flaws in the cycle :wink:

#425

Thanks for the input Stephen - I’ll see if I can get a bit more extension in there. The problem I’m having isn’t jitters - it’s joint popping. I tend to set my bones up pre-bent, and when they go perfectly straight and then suddenly bend back again, there’s usually a pronounced “pop” in the joint. It has to do with my rig, I’m sure. It seems like the angular acceleration coming away from being perfectly straight is a bit excessive… sigh.

I may have to revisit the rig - see if there’s some way I can add a joint limitation of some sort. I wish I could figure out a way to do a distance-based constraint between two bones - thus keeping the hip and heel from getting to the point where they go perfectly straight - just a tiny bend…

Thanks again for the input and the reference :wink: - I’ll post it if I can figure out a way around it - and get those legs just a tiny bit straighter…

JoeW

#426

Okay - I ended up pretty much re-doing the whole darn thing - LOL… I went in to make some changes - and ended up touching everything… I can already tell that the “practice” is doing me a lot of good. After walking around a while and really paying attention, I realized that the first animation had too much height on the passing foot - making it look like she was marching or stomping. I also realized why we (humans) trip so easily…

Here’s the new improved walk cycle - with some upper body added for good measure… maybe tomorrow I’ll get to the arms and head… sheesh - amazing how much time 36 frames of animation can take :slight_smile:

Thanks again, Stephen - your suggestion of “two frames of straight leg” came in handy :wink:

http://www.hash.com/users/joewllms/cgtalk/WalkTest6-27.mov

After this is finished, I’m going to try a couple of different walks (and a run cycle) - something more sensuous, and something a little more relaxed… It’s time consuming, but adding “character” to the walk is a great exercise in observation…

JoeW

#427

Nice one Joe, the walk has really improved, which is amazing considering there wasn’t much wrong with it before.
I’m still gob-smacked at how gorgeous this model is, although her feet seemed to have grown :slight_smile: Your channel wrangling must be very good as there’s not the tiniest visible jump between cycles. I’ve heard people say just ‘zero slope’ the end keyframes. I don’t, what I do is spend inordinate amounts of time getting the end and beginning frames channel graph lines to follow the same trajectory. Do you have any tips?

#428

Thanks Stephen - your mild criticism (constructive) help spur me on. I’m a lot happier with this iteration.

Her feet seem larger because of the context - if you saw her with her head and headress, as well as arms and hands, it makes her feet look proportional. When I built her, I struggled with how large to make her hands and feet - didn’t want too dainty, and didn’t want too huge - and I didn’t make final adjustments until the model was complete, and I could judge proportions based on the whole package :wink:

As far as getting the curves to line up, I use zero slope only where it makes sense - but as you know, there are places where something is accelerating or decelerating at the end of the cycle. Starting frames can be tough, too because it’s hard to match the mag and gamma of the exiting spline on the other end. I decided to use a trick I learned in Lightwave - I copy keyframes into negative frames, as well as frames beyond the end of the cycle, thus eliminating the guess work of mag and gamma tweaking. You can’t copy keyframes into negative frames (per se) but you can create a keyframe (click on the dotted channel line) and then place it and change it’s settings. It’s been so long since I’ve really tried any serious animation - I’m going to have a few suggestions for the Hash guys. One thing I’d like to see is easier “looping” of channels (more like other programs do it), as well as a “maximum distance between bones” constraint. More stability when editing channels would be nice, too - I’ve been getting really good at making AM just “go away”… hones my “save discipline” to a fine edge…

JoeW

#429

Oh, don’t get me wrong, I like the size of her feet, they are a perfect fit, she is indeed a lady of ample proportions. Part of my day job involves drawing Disney characters and it’s always pointed out that I’ve drawn Cinderella’s feet too big, so maybe I’ve become a little neurotic where women and feet are concerned.
Putting keyframes temporarily into the negative, is an excellent idea…cheers.
Not sure what ‘looping of channels’ means, it certainly sounds useful. One thing I wish Hash would implement, is at least one undo command in the channels, it’s really tedious being forced to revert instead :wink:

#430

Ah - drawing Disney Characters - man, I’m such a klutz with a pencil these days… I gotta choose my battles, though :slight_smile:

Actually, I go ahead and leave the keyframes in the negative and the positive areas - you can just specify what to crop when you use the action in the Cho. Saves having to guess what will happen when you delete those keyframes… which can be scary sometimes… It would be nice to be able to specify how long an Action is in the Action window as opposed to having to crop it in the Choreography, though…

Looping (repeating) means that you set an action as a cycle, and then your channels are “ghosted” to show what they look like if they’re going to be repeated. You can see this if you look at http://www.hash.com/users/joewllms/cgtalk/Channels.jpg

JoeW

#431

Hello.

I could be wrong (it wouldn’t be the first time) but I believe Hash already has this looping feature. They call it “Accumulate” though. Check out the bottom of this page.
http://www.hash.com/am2002/Time%20Line/index.htm

Stephen: I’m relieved to hear that Briar is doing okay, and that Bob’s work cured that jitterbug problem. Please show us more stuff! I need to be convinced that everything is okay! :wink:

Joe: The new walk cycle is better. I didn’t think it could be improved upon. The changes are so subtle, and yet it’s undeniably better.

Sincerely,

Carl Raillard

#432

On second thought, “Accumulate” is a bit different from “Looping.” However, you might be able to get the Accumulate feature to work like Looping I think if the start and end keyframes were identical.

If I’m wrong, correct me.

Sincerely,

Carl Raillard

#433

Hey Carl - Looks like you are correct - there is a “repeat” function on the in and out side of the keyframes - this should help a HUGE amount - thanks for pointing it out, and thanks for the compliments on Hunter’s walk :slight_smile:

I also found that you can change your timing, AND slide the frames around - this is AWESOME - thanks again! :thumbsup:

JoeW

#434

Great! I’m glad I could help!

Joe, you also mentioned something about a new “maximum distance between bones” constraint. I’ve been mulling over an idea that might be similar. I’d like your opinion of it. Here’s a link to a picture that I’ve sketched, that illustrates how this new constraint might work.

http://www.sonic.net/raillard/hash/spherical_translate_limits_constraint.gif

Is this what you had in mind? Or did you have something else in mind?

Sincerely,

Carl Raillard

PS: I’ve just posted add’l info about this new constraint on the Feature Requests thread.

#435

Hi Carl,

Yep -that’s pretty much what I had in mind - with the possible inclusion of making it so that you could set the distance to be from the parent’s root to the child’s tip. I know there are “rotational limits” that are supposed to do that kind of thing, but I’ve never had any luck getting them to work with other constraints - they tend to get a bit wiggy at the wrong time…

Thanks for posting that - I know that the more they hear a request, the more likely it is to get implemented…

JoeW

#436

Wow! Thanks!

I’ll get to this as soon as I get home (about 2 days).

Zack

#437

Ok, I redid the lighting. It’s nowhere near finished, but as I’m leaving (again) tomorrow (this time for Portland, OR) I decided to post a milestone.

There are a whole bunch of lights, the bright side is that it now has excellent antialiasing. The unfortunate part is that it’s still lacking definition, and it now looks a bit washed out.

Thanks for the help,
Zack

guitar0.jpg845x839

#438

Hi zero2zillion,

Coming on nicely,

I am thinking that maybe you could put some kliegs in there to give it more lighting variation around the image. At the moment everything is very evenly lit.

Texture on the wall will help a lot also.

If it was me making it, I would brek up the verticle of the corner of the wall with the guitar by having the guitar on more of an angle, moving it to the right and bringing the camera round to the left.

Also, the colour of the body is maybe a little dark, maybe lighten the colour or get more speculars in there.

Hope the above helps,

Good luck

#439

Hi, I’m in Portland I discovered this great bookstore called Powell’s that had tons of new and used 3D related books, including used animation master handbooks (I didn’t buy the handbooks though, one was the original Jeff Paries one, the other was one titled “Animation” and must date back to '90 or earlier). I found a used version of Jeremy Birn’s [digital] Lighting & Rendering and it’s really informative.

I’m probably going to redo my lighting based on this as soon as I get a chance (could be a month…)

#440

Ok, here is the updated lighting. You’ll notice that the bottom is also more reflective. The main problem right now is the headstock material.

Critique Away!

guitar.jpg845x839