I have given up on my "fixed" pose, at least for the moment. I wanted to see if the thighLength, feetDistance, and hipHeight parameters in the pose were making any difference to the WD output. So I did this test.

The wire frame is the standard P6 (not P4) walk. The solid figure is is the same walk, but using Antonia specific parameters, namely:

thighLength 0.175231
feetDistance 0.079002
hipHeight 0.381689

My new foot geometry was used in the figure, but, unlike previous tests, limits were turned off.

As you can see there is a considerable difference the separation of the feet, presumably due to different "feetDistance". In side view there has also been an effect on the amount of xrot (kick) applied to the thighs, presumably due to "thighLength". "hipHeight" on the other hand, does not seem to have had any visible affect on the altitude of the hip. However the ytran values are slightly different.

-0.012452 for standard P6 blend,
versus
-0.013336
For Antonia specific

The default P6 values seems to give better results than the Antonia specific values, in this test. Perhaps the important thing to note is that we appear to be able to tweak WD output by changing these values.

Interesting.  Is it handling better without limits, now that the alternate foot geometry is in use?

I'm curious about thighLength, particularly.  The other two are obviously used by the Walk Designer only, but thighLength is included in all poses.  How is Poser using it, outside of the WD?

Quote - Is it handling better without limits, now that the alternate foot geometry is in use?

No, but I did not want the limits interfering with the changed parameters, that might have muddied the waters, and made the results of the test harder to interprite.

Quote - I'm curious about thighLength, particularly.  The other two are obviously used by the Walk Designer only, but thighLength is included in all poses.  How is Poser using it, outside of the WD?

I don't think it is used outside of the Walk Designer. It's being written into the file in case the file is to be used in the WD. "thighLength" has been around since at least P4. I don't think feetDistance, and hipHeight existed in the original P4 version of the files that shipped with P4.

The image above uses the default P6 walk blend with Antonia parameters, except for hipHeight, which has been set to 0.100000.

thighLength 0.175231
feetDistance 0.0652754
hipHeight 0.100000

Ooh!  So if we need her thighs to bend less, we can raise that value.  That might help with the knee-buckling problem.

That sounds like a reasonable assumption. In your AP4 walk.pz2 we have "hipHeight 0.381689", I will try raising that value to 0.400000, and to 0.430000, and see what happens.

This could have other concequences, like shortening the stride, making the feet slide, or making the legs bend less at the knees, which might make the walk look stiff. We will just have to see.

I think I'm going to have to start looking at feet sliding, have you noticed any so far with your AP4 poses?

Did you get the updated walk cycles I made yesterday?  I think I've already set Antonia's hipHeight at 0.4 in those. 

Two completely standard instances of Antonia-1.0.0, no special geometry nothing done to the limits. Using your "AP4 walk" (the original one, not the new one).

The figure on your left uses the unaltered pose with "hipHeight  0.381689".

The figure on your right uses "hipHeight  0.430000".

Quote - Did you get the updated walk cycles I made yesterday?  I think I've already set Antonia's hipHeight at 0.4 in those.

I did get them, but must admit, I have not tested them yet.

That's looking pretty good.  There's still a bit of knee-buckling, or at least an abrupt bending of the knee in the middle of the animation.  How is it looking, in terms of feet penetrating the floor?

The newer walk set integrates some of the changes we discussed earlier, but it's no biggie.

Is there anything I can test, with the walk process, right now?

Here is exactly the same thing from the side. Original hipHight in white outline, greater hipHight in solid figure.

Quote - That's looking pretty good.  There's still a bit of knee-buckling, or at least an abrupt bending of the knee in the middle of the animation.  How is it looking, in terms of feet penetrating the floor?

Raising the hip has not helped the knee bukle, in fact, contrary to our expectations, it seems to have made it worse, though that will be hard for you to see in these small animations.

Quote - Is there anything I can test, with the walk process, right now?

As I feel like I'm walking through trekle in the dark, hoping to bump into something useful. I think you should follow your own intuition as to what to test. However there is something you mightbe able to do, if you feel like it. In your "AP4 walk", if you could mirror the values from the left leg, in frames 16 through 30, over to the right leg in frames 1 through 15. Then I might be able to get a walk without the toes going below the ground, though I would still need to use the injected geometry to achive this. Only if you feel like it. I tried but keep messing it up.

At the moment I can get the left toes to stay above the ground, but not the right toes, the right toes are still a problem in frames 7 through 11.

Do you want to have both legs mirrored this way?  That is, can the poses simply be flipped, so frames 1-15 are the same as 16-30, but flipped left ro right?  Or just the one leg?

Frames 7 through 11 in the right leg are "broken", meaning they don't work OK under my system. So I don't want the "broken" right leg mirrored to the good left leg, or at least not frames 7 through 11, where the problem is.

Cage,

Some good news and some bad news. The AP4 walk_flip3.pz2 pose that I requested, and that you kindly made for me, does, as I had hoped, prevent the toes going through the floor, when used in combination with my geometry injection for the feet.

Now for the bad news. Whilst the problem with the toes has been fixed, there are now some new problems. The WD wants to create a much smaller number of frames for this new walk blend, 84, as opposed to 158 for the standard "AP4 walk" (old version). If I accept this number the figure strides ahead of her twin who is using the old walk. Manually setting the figure to use 158 frames for the walk, does not dampen her desire to stride out ahead of her twin. There are also problems with the right leg flying out too far behind in some frames, and the feet not being in contact with the ground when they should be.

It's early days, and I'm just reporting my first results, I will get back to you when I know more. At first sight, it seems we have solved one problem, only to find we have created several others.

Hmm.  Very strange. 

All I changed for the pose was the keyframing for the trans and rot dials for the right thigh, right shin, and right foot.  Possibly the translation keyframing could create some issues, but it looked like there were already keys for each frame for the trans dials of all actors in the pose.

If the stride length of the right leg was increased by mirroring from the left side, that would explain the discrepancy in the number of frames, perhaps.  I'm more toubled by the leg flying out behind.

Something I didn't do was adjust the hip rotation in the altered frames.  Possibly that should mirror the hip rotation in the frames from which we've borrowed?

Hmm.  Looking at the edited pose with the mirroring, it looks like the rotation order for the right foot is different.  I pasted in the whole actor blocks from a pose I saved out of Poser.  Poser saved the Antonia pose with a different rotation order.

Which may raise some questions.  Does the order of the rotation parameters in a pose affect anything?  Do the rotation orders in our Posette source poses match those of Antonia?  Possibly we can improve results by making sure the orders are correct for Antonia.  Some of the trouble we're seeing could be strange euler rotation problems because Poser's application or an ordered rotation is being foiled by incorrect poses.

I finally got round to doing a comparison between your "AP4 walk", old and new versions (nothing to do with walk_flip3). This is the output, after using them as walk blends at 100% in the WD. Seen from the front you have to look very hard to pick any difference. Seen from the side, there is more difference, the whole figure is shifted back in the new version, and there are differences in the amount of thigh and shin xrot, but not a lot. Personally I could not pick one as being better than the other.

Solid figure is the new version, yellow outline is the old version.

I will get back to walk_flip3 in my next post.

Quote - All I changed for the pose was the keyframing for the trans and rot dials for the right thigh, right shin, and right foot.  Possibly the translation keyframing could create some issues, but it looked like there were already keys for each frame for the trans dials of all actors in the pose.

The only actors that should (possibly) have none zero translations are the hip, feet, and hands. I have checked, all the translations are the same in flip3, and AP4 walk.

Now this is interesting. For the right foot the translations also seem the match between files. But in both files, all the translations are zero until frame k 15.

This does not happen in the left foot, in the left foot all translations have none zero values. This does not tell us anything specific about flip3, but it probably explains why I am seeing the right toe going under the ground, but not the left. If I am right, it's an important discovery.

Quote - Which may raise some questions.  Does the order of the rotation parameters in a pose affect anything?

I don't think the rotation order makes any diffrence in a pose file, I think the figure applies them in the order they are in in the cr2, irrispective of the pose. I just think that, I have never done any tests to prove it, and what goes for a pose used as a normal pose, may not apply to a pose used in the WD.

Quote - Now this is interesting. For the right foot the translations also seem the match between files. But in both files, all the translations are zero until frame k 15.

This does not happen in the left foot, in the left foot all translations have none zero values. This does not tell us anything specific about flip3, but it probably explains why I am seeing the right toe going under the ground, but not the left. If I am right, it's an important discovery.

That is interesting, and it sounds important!  If the WD is operating as though IK were on, foot translations could make a great deal of difference.

Quote - I don't think the rotation order makes any diffrence in a pose file, I think the figure applies them in the order they are in in the cr2, irrispective of the pose.

I'm not sure.    When I was trying to write a Poser pose exporter for Blender, the results differed when I switched the rotation orders.  But that could have had less to do with anything on the Poser end than on the Blender end, where the euler ordering may have been different with the changed order.  Hmm.

Hmmmm.... the rotation order is what you need to write down for Puppet Master, the program creates a txt-file containing the order but you have to look through it (the script will remind you to do this so it's not a bug) and there is always something that needs fixing.

As the walk designer is as well a kind of pose-converter it may be necessary to tell it the rotation order (somewhere). I've looked into the Walk designer folder in the Pose Folder and found a "Don for Walk Designer.pwd". It's not plain text so so far I am not sure what it contains. So I am not sure if this is the file that may contain the rotation order. However: If the rotation order is of any need for you let me know and I upload a txt file that contains it.

The .pwd file is a bit of a mystery.  I assume it's a new binary file type used internally by Poser.  It's probably just what we need in order to customize a wilk to a figure, too.  :lol:

I think Les is probably right, and the order in which rotations are presented in a pose doesn't make any difference.  Although I haven't tested that.

You need to use the old Antonia 1.0 (or any version prior to 1.2), which should still be available at the Free Site.

Yes, Cage is right. I use Antonia 1.0 to create the poses and then run the script we are talking about in the Opinions-Thread. I kept a copy of all 1.0 poses anyway at the moment, just for the case (whatever case this may be, but I am a kind of hamster when it comes to WIP).