It looks like the problem with the feet penetrating the ground may be inherited from (at least some of) the P4 pose files. 

Any adjustment to how far apart the legs are is a fairly easy tweak, modifying the seed pose from which the bias values for the script are being derived.  Currently I've tried to line up the innermost points of Antonia's feet with those on Posette.  That ignores any difference in alignment of the legs, however, so it could create trouble in the knee area.  Ideally, the joint center for Antonia's feet should be aligned as well as possible with the joint center for Posette's feet.  I'll try setting that up.

Beside the too poking through the ground this looks amazing, lesbentley! Especially the way how arms and shoulders move is very natural!

AP4 walk, limits on.

Here is a side view of frame 11, the right toe is way below the ground. I also notice that the hip is zTranslated out in front of the BODY by 0.0583265 in all frames, this value seems to come from the pose, though slightly modified by the WD. In the pose the value is 0.0583264993891. In the image, the rear cross is directly below the BODY origin, and the front cross direct below the hip. I wonder if setting the value of the hip ztran to zero in the pose would have any affect on how the feet work?

SaintFox,

Everything above the thighs, in my previous image, is down to Cage's "AP4 walk.pz2". All I did was to apply some limits to the legs. It's a small easy tweak, but has a big effect on the legs.

That's a fairly easy pose edit, to test the idea.  Just delete the Z Tran keyframs for the actor in a copy of the file.

I'll try to get some adjusted poses together in a bit.

The poses I'm producing are created using a version of the correction script which Les requested.  So LB should get credit for any progress here.  :laugh:

I had a theory that the toes going below the ground was due to the length of Antonia's feet (the actual foot actor is short because of the instep). I thought that the WD was rotating the foot untill the front edge touched the ground. in some tests it looks like this theory had to be right, but in others it looks like it has to be wrong.

Quote - @odf:  If you're there, what do you think about adding an IK-priming pose to the main distribution when version 1.0.1 comes out? 

You mean version 1.2.0?

There is a pose called "Standard" by phantom3D a.k.a. Diogenes in the distro that does exactly that. It used to be Antonia's default pose back in the day, but I decided it was better to load her all zeroed out. Maybe I should rename this pose to something more obvious. How about "IK-friendly"?

Quote - The poses I'm producing are created using a version of the correction script which Les requested.  So LB should get credit for any progress here.

That's very gratious of you Cage, and if I had written the script it would also be true. All I did was have an idea, it's you who did the work to put it into practise. I also have an idea for world peace, now if I can find someone to put that into practice!

"IK-friendly" sounds good to me!

Quote - I had a theory that the toes going below the ground was due to the length of Antonia's feet (the actual foot actor is short because of the instep). I thought that the WD was rotating the foot untill the front edge touched the ground. in some tests it looks like this theory had to be right, but in others it looks like it has to be wrong.

I thought it was because of the instep actor. If WD assumed the feet to be rigid, it would place them with their lowest points on the ground and further geometry attached at their ends would then be below ground in some positions.

I tried zeroing the hip ztran, whilst it did have an effect on the way the WD applied the walk cycle, it did not cure the problem. :sad:

Whilst the toes are going slightly below the ground in the pose in some frames, especially fremes 28 through 30, It's the WD that's adding the vast bulk of the problem!

Quote - I thought it was because of the instep actor. If WD assumed the feet to be rigid, it would place them with their lowest points on the ground and further geometry attached at their ends would then be below ground in some positions.

Exactly my thoughts. However I seem to have seen situations where this is not happening, or at least not happening in all frames. It's very confusing. I think the length of the foot geometry, or the position of its end point does play a part. But there also seem to be other factors which I don't understand. I seem to remember that injecting a longer foot geometry did help, in tests I did some months ago, but I don't think it cured the problem completely. Memory is a bit fuzzy on that though. Perhaps I should try that again, but its an inconvenience for the end user.

I think the ideal solution, aestetically, would be to bend the instep and toe to make up for the ground penetration. But I don't know whether that's feasible technically. Could it be included in the source walk cycles?

Okay.  I think I understand what's happening with the toes going through the floor.  That doesn't happen with "walk in place".  It's being applied by the Walk Designer, when a path is used.  It looks like the WD is assuming that the ball of the foot will be part of the Foot actor.  Watch how it happens.  The toe and instep actors go through the floor and the penetration stops at the front of Antonia's Foot.  She then puts her weight at the front of her foot and pivots.

I think we have a disagreement between the actor divisions in the geometry and the expectations of the WD.  The only way to fix this may be to have the feet cut differently on a WD-compatible Antonia.

Assuming I'm not completely wrong again.  :lol:

Quote - You mean version 1.2.0?
There is a pose called "Standard" by phantom3D a.k.a. Diogenes in the distro that does exactly that. It used to be Antonia's default pose back in the day, but I decided it was better to load her all zeroed out. Maybe I should rename this pose to something more obvious. How about "IK-friendly"?

Ooh!  Great!  From the thumbnail, I thought it was a copy of the "Attention" pose, with a different name. 

And, yes.  Antonia 1.2.0, now that that's been decided.  :woot:

Y'know, if you release it on Saturday, that will make it special for Cage.  :b_funny:

(Fixed a typo which changed meanings disastrously)

WOW! That's a huge improvement!:biggrin:  Frame 11, exactly the same set up as previous image, but with new foot geometry. White outline is old position of foot. The new geometry stretches one vertex in the foot out the the start of the toe actor.

For the left foot this is close to a complete cure, however the right foot, though better, is still going badly below the ground in some fremes. Why should it fix one foot, but not the other. Is it something in the pose?

And I'm still not very happy about asking the user to inject new geometry. If there is a better solution, I would like to find it.

Quote - For the left foot this is close to a complete cure, however the right foot, though better, is still going badly below the ground in some fremes. Why should it fix one foot, but not the other. Is it something in the pose?

And I'm still not very happy about asking the user to inject new geometry. If there is a better solution, I would like to find it.

You're testing the new poses?

The bias pose I'm using is symmetrical, so if the problem is coming from the poses, it's there in the P4 original.

If we can come up with a solution requiring no geometry changes, that would be good.  My theory, though, is that the Walk Designer will fight our efforts until we give it a Foot actor which includes the ball of the foot.  I don't think we'd need to inject geometry.  Odf writes about distributing a stripped-down version for the Walk Designer.  We could ship just a low-res Antonia, but one with the feet re-cut, or something.  :unsure:

Quote - Y'know, if you release it on Saturday, that will make it special for Cage.  :b_funny

Why? Is that your birthday?

Quote - I don't think we'd need to inject geometry.  Odf writes about distributing a stripped-down version for the Walk Designer.  We could ship just a low-res Antonia, but one with the feet re-cut, or something.  :unsure:

Ideally, users would just call Walk Designer on the default figure, and get out their animations without further ado. But if that's not possible, using a special WD figure to create the walks and then some post-processing to make it work with the "normal" Antonia does not seem too bad to me. Cleaner than injecting geometry, anyway, if that's the alternative.

If a special figure is used and the instep actors remain rigid, maybe it's possible to just combine foot and instep for this "WD-Antonia", so that no point pushing or extra adjustment will be needed? If I understand correctly, the output of WD will simply be an animated pose, so if the instep actor is not mentioned in that, no harm will be done.

"AP4 walk" with limits set and new geom. Here is a are 29 frames of the walk cycle on a straight path (29 loops better than 30). As you can see, the left foot is fairly good, but the right foot still goes below the ground in places. The new geom helps a lot , but is not a complete cure.

Weird! I'd really like to know where those strange assymetries come from.

Has anyone tried this kind of thing with stock figures? If the same problems occur on P4, at least we know it's not entirely "our fault".

Quote - If a special figure is used and the instep actors remain rigid, maybe it's possible to just combine foot and instep for this "WD-Antonia", so that no point pushing or extra adjustment will be needed? If I understand correctly, the output of WD will simply be an animated pose, so if the instep actor is not mentioned in that, no harm will be done.

That's correct.  We'd generate an animation which will have the actors re-named, for use of Proper Antonia.

If the instep actors include the right portion of the foot geometry and my guess about what's happening is correct, merging the foot and instep seems like it would do the trick.

But we'll have to distribute these poses somehow, too.  I don't think we can distribute poses which are derived from the P4 walk cycles, can we?  If not, we'll need to encode them,  or something.  But I notice that Poser 8 doesn't include the P4 animations.  So how can we distribute these walk cycles? 

Quote - Has anyone tried this kind of thing with stock figures? If the same problems occur on P4, at least we know it's not entirely "our fault".

I think that's the next logical step.  I'll try it, but I'm not as observant as Les is, with these things.  :lol:  I was ready to decide it was good enough a bit ago, when it clearly wasn't.  Cough.

Quote - "AP4 walk" with limits set and new geom. Here is a are 29 frames of the walk cycle on a straight path (29 loops better than 30). As you can see, the left foot is fairly good, but the right foot still goes below the ground in places. The new geom helps a lot , but is not a complete cure.

Overlooked this, due to cross-posting.

Are you using a walk path?  Does this show up in the walk-in-place, or when the walk is applied as a regular pose? 

Quote - I think that's the next logical step.  I'll try it, but I'm not as observant as Les is, with these things.  :lol:  I was ready to decide it was good enough a bit ago, when it clearly wasn't.  Cough.

Honestly, whenever I get a notification that Les post in the main thread, I panic a little, because he tends to spot all those little glitches that I've never noticed or hoped I could get away with. Of course, I'm always glad afterwards that he did.

Quote - Of course, I'm always glad afterwards that he did.

Absolutely.  I wish I was so observant.  :thumbupboth:  I'm laughing at myself, because I'm so hopeless in that area.  :lol:

Quote - Ideally, users would just call Walk Designer on the default figure, and get out their animations without further ado. But if that's not possible, using a special WD figure to create the walks and then some post-processing to make it work with the "normal" Antonia does not seem too bad to me. Cleaner than injecting geometry, anyway, if that's the alternative.

Another whole new version just for the the WD "Cleaner than injecting geometry", not too sure about that, how many versions are we going to end up with? Sounds like it could get confusing.

Quote - maybe it's possible to just combine foot and instep for this "WD-Antonia".

Would it be possible to have one orphan vertex out near the start of the toes? Personally I like that idea better, but don't know if it is possible to have an orphan vertex, and if it is, if the WD would recognise it.

Quote - If I understand correctly, the output of WD will simply be an animated pose

Not exactly. The output of the WD is an animated figure in the Document Window. You could then save an animated pose from that figure. But this would create a LOT of extra steps for the end user, and I don't think it is a viable solution.

In that system, the user would have to load a new version of Antonia into the scene, might have to hide the old version, which would be superimposed, then run the new version through the WD, then save a pose, then hide the new version and show the old version, then apply the pose. That's just the simple scenario, assuming that the walk cycle is not being applied part way through an existing animation, and assuming that the user has P7 or later, because an animated pose won't save the BODY translations in P6. No that's not a viable solution.

Using a special figure to produce walk blends for use in the WD is OK, because you only have to do that once, then you can distribute the blend poses. The blends can then be used to create numerous specific walk cycles. Asking the end user to use a special figure to produce a specific walk cycle, then transfer the results to another figure, is not an acceptable option, in my opinion.

Quote - Has anyone tried this kind of thing with stock figures? If the same problems occur on P4, at least we know it's not entirely "our fault".

I don't think the assymetries are our fault. The poses used in the WD have assymetries built in to them. Cages poses (don't know bout the new set) are based on the P4 poses, and probably contain the same assymetries.

Quote - Are you using a walk path?  Does this show up in the walk-in-place, or when the walk is applied as a regular pose?

Using a path. Have no interest in Walk in Place, unless as a step towards following a path.

Quote - Using a special figure to produce walk blends for use in the WD is OK, because you only have to do that once, then you can distribute the blend poses. The blends can then be used to create numerous specific walk cycles. Asking the end user to use a special figure to produce a specific walk cycle, then transfer the results to another figure, is not an acceptable option, in my opinion.

You may have a point there. But is it really necessary to have the WD-Antonia loaded into Poser in order to create walks for her? Maybe I remember it wrongly, but I thought one told WD which figure to use as a template (via the "figure type" button or whatever it's called) and it then created the walk from that, to be applied to the selected figure. So the only problem then would be that WD would try to pose the "hip2" actor, which does not exist in the standard Antona.

If we made the "new" hip (formerly hip2) the root of the actor hierarchy in WD-Antonia like it is in the P8 figures and then fixed the generated animation via a post-processing script, might that work? Or alternative, switch the actor names hip and waist in WD-Antonia and switch it back in post?

In the pose the toe does go slightly below the ground in frames 28 through 30. Note that frames 28 through 30 are cured in my last animation, but now there is a problem with the right foot in frames 9 through 11 in the animation (though they are improved) . I don't know that happens with Walk in Place I have not tested that.

Quote - I thought one told WD which figure to use as a template (via the "figure type" button or whatever it's called).

I have always suspected that the "figure type" button only sets the figure to be used in the preview window in the WD. But I may be wrong.

Quote - You may have a point there. But is it really necessary to have the WD-Antonia loaded into Poser in order to create walks for her?

The WD can only output a walk into a figure loaded in the current document. In the WD you get a list of figures in the document, and must choose one of those to apply the walk to.

Quote - > Quote - You may have a point there. But is it really necessary to have the WD-Antonia loaded into Poser in order to create walks for her?

The WD can only output a walk into a figure loaded in the current document. In the WD you get a list of figures in the document, and must choose one of those to apply the walk to.

Sorry, I phrased that wrongly. What I meant is, could you select WD-Antonia as the figure type and then apply the walk to default Antonia. I did a quick test a while back, using Alyson as the figure type in WD (but applying the walk to Antonia), and that seemed to produce better results with the built-in blends than picking Antonia.

By doing what odf suggested, and the thing with the IK, I was getting a walk without going throught the floor, but not as nice a walk as les has been getting.

Quote - Sorry, I phrased that wrongly. What I meant is, could you select WD-Antonia as the figure type and then apply the walk to default Antonia. I did a quick test a while back, using Alyson as the figure type in WD (but applying the walk to Antonia), and that seemed to produce better results with the built-in blends than picking Antonia.

I don't know the answer, but it's got to be worth a try.

@Les:  I just tried applying your leg-limits pose.  In the three walk animations I've tested since loading it, Antonia steadily rises into the air as the walk progresses.  It's kind of surreal, actually.  :blink:

Anyway, I'll try it again.  Something weird is happening.

@odf:  I just tried your suggestion of animating Antonia 1.2 but selecting Antonia-WD as figure type.  This seems to generate walk cycles in which the Hip is not moving independently of the upper body.  (However, this test overlaps with the one mentioned above, so things are somewhat in need of sorting out, here....) 

Quote - @Les:  I just tried applying your leg-limits pose.  In the three walk animations I've tested since loading it, Antonia steadily rises into the air as the walk progresses.  It's kind of surreal, actually. :blink:

Anyway, I'll try it again.  Something weird is happening.

Have you tried attaching weights to her feet? :lol: :lol: :lol:

Seriously though. I had that happen once, but in a totally difrent context. I think one figure was parented to another when it hapened, or perhaps to a prop. I have not had any levitation experiences with Antonia.

P.S.

Sorry about the levitation crack, I had failed to comprehend the gravity of your predicament! :lol:

Very strange.  It's like the Z translation has been linked to the Y translation.  :scared:

News flash:

In a statement released early this morning, a NASA spokesperson, denied reports that a Poser scientist had beat them to the invention of the first practical anti-gravity drive. The spokesperson said that Professor Cage's invention was a hoax, and could not possibly work.

Quote - P.S.

Sorry about the levitation crack, I had failed to comprehend the gravity of your predicament! :lol:

:laugh: :lol:

Cage,

Perhaps the WD thinks that the knees are bent too much, or that the feet are too low, and is trying to raise the hip to compensate?

You guys crack me up!

Quote - Cage,

Perhaps the WD thinks that the knees are bent too much, or that the feet are too low, and is trying to raise the hip to compensate?

If that's what was happening, it just kept going and going.  :scared:  I didn't save the file and the results didn't recur when I applied your pose to a freshly-loaded Antonia.

Something really weird happened, though.  I'm not making this up!  :lol:  It might happen to you, some day, when you least expect it.

As I like to say, Poser has the gremlins!  :lol:

This is the WD output from Cage's "AP4 walk", tweaked by hand after it came out of the WD, to stop the toes going through the floor. Next I'm going to feed it back into the WD as a walk blend, and produce a new walk cycle from it. I am not particularly hopeful, I predict that the WD will mess it up.

My "fixed" pose won't work in the WD, when I try to apply it the slider just resets to zero. :crying:

I'm going to bed. Had enough for one day.

Quote - My "fixed" pose won't work in the WD, when I try to apply it the slider just resets to zero.

How did you create the fixed pose?    You might need to edit the hipHeight, thighLength, and feetDistance back into a library-saved pose, in order to make it Walk Designer compatible.  I'm not sure whether there's anything else about such poses which differentiates them from regular poses.  :unsure:

Quote - How did you create the fixed pose?

Just moving the feet and hip around in Poser. I also played around in the Graph Editor, so I guess it's possible I did something to the spline type settings, or lost some key frames.

You saved it to the library, I assume?  Have you looked at it in a .cr2 editor?

Quote - You saved it to the library, I assume?  Have you looked at it in a .cr2 editor?

Yes, and yes. I have made a new version. Some of the fremes were not key framed, so I fixed that. Deleted the centre of mass stuff that Poser had added. Included Morph channels, because I think the P4 files do that, added the feetDistance, hipHeight stuff, deleted some junk from the end of the file. Still not working. Yet I have made poses in the past that worked in the WD. Very strenge.

Maybe you could try running ExamDiff or something on your new pose, to compare it to an existing one and see what could be causing the trouble.  If it comes down to it, a script can be created to port your changes to an existing walk pose which does work.

Poser.  Gremlins.  :unsure: