Report about new MHX (mix 2 rig way)

[engetudouiti]

I feel, when you up-date, there seems some un-necessary pos (and rot) change…without your intention.

There are 2 issues, (I suppose so)

1. bend.L and twist.L (of arms) need to locate on the fk.L bone (cupsel bone) . correctly　

or FK.L bone need to locate on, the bend.l and twist.L to cupselize 2 bones. (or snap may change I afraid)

2. when generate twk bones, it can be locate on same place, as org bone, I suppose. then twk bone will be real deform bone.

org bones only used to keep def daz rig. (at current it use to deform too, but I suppose you need not, but use twk bones,

twk bones simply generated as duplicate org bones, and use it to deform (with constrain org bones)

（Do I need to re-export duf? if so I try again)

[engetudouiti]

Yes only test with Zero pose, then just snap IK to FK (though it already IK, I test it)

it change deform that means, zero pose IK pos and FK pos not locate on same place.

You need to locate, Cupsale FK <> IK bone, and each bend, twist bone, on to same place as one line, and keep same role. without any delta . (it is MHX snap rig )

[engetudouiti]

What you need was,, (I suppose)

1. return version once.. (which not add new thing) but remove your double Shin.

because old arrangement show perfect bone location for FK <> IK (Cuplsale pose bone) and divided 2 bones for bend and twist. then remove current twk bones. (because it try to copy MHX rig bone location) with of-set.

2. generate new def rig bone chain. which will use constrain, and transfer pose from FK <> IK MHX chain

3. generate new twk bones, which simply duplicate all def rig bones. and set constrain. with off-set. and use it as real deform bone.

So it can still transform as user need as same as before,

‌

[Thomas Larsson]

I don’t think that the new deform bones affect fk-ik snapping in any way; that is done entirely within the control rig. There are problems with snapping, but those were already present before. I just tested with stable 1.6.1. At rest, the fk pose is the same as rest pose, but the ik pose is slightly different.

‌

[engetudouiti]

Do you only test with G8? I only test with G3, then I clear see, FK arm and IK arms not generate as same pos for rest pose.

(and when convert as IK,, it start as rest pose.. then as you see my pic,

IK or FK cupsel bone, not correclty locate on to bend and twist bone.

Is there any reason, you make it so? about old version, with out bone limit, and auto twist. I do not see any snap issue about G3. because FK and IK chian (cupsel) bones should locate on to the bend, and twist bone. with show same bone roll. = so snap should work perfect.

(but to snap work, we need to off limit , and can not use follow twist with hand bone, it never work)

if I do not need to re-export duf, I may hope you test with def G3 first. then after generate MHX, try snap IK to FK , if FK, IK cupsel controll bone, and bend twist bone (it need to divide bone deform ) are generated correctly, it never show any deform. (but I see it deform hard, when IK bone snap to FK, with rest pose (IK pose)

I may show pic again,, with zero pose, how each IK FK (and bend twist bone) generated with zero pose. (MHX)

[engetudouiti]

So now I just convert to MHX >> I expect it show zero pose . (I do not import any morph)

then change mode from pose to edit, now it deform hardly.

I do not think it is usual behavior. (I do not see such issue before) for G3.

‌

[engetudouiti]

And with new version,,

it show flip twist issue for me..

‌

[engetudouiti]

  <div class="preview-container wiki-content"><!-- loaded via ajax --></div>
  <div class="mask"></div>
</div>

</div> </form>

[Thomas Larsson]

I replaced the copy transforms constraint with copy rotation, and that seems to fix the flip issue on the upper arm. The forearm can still deform badly, but that only happens if that hand rotation is outside the limits.

[engetudouiti]

The problem is, you seems change elbow pt ik and elbow link ik bone setting somehow.

Now I edit (manually with eye) and correct it, then Edit mode <> Pose mode not show deform anymore.

You need to use math and how locate those bone and roll with check length.

I had said it and you did it before ^^; (I believe so) when I suggest how to snap IK FK correcrtly for MHX, but with up-date versions, you seems change way I suppose.

to get good IK roll we need to keep angle for twist and bend. it make triangle.

then you set target and bone roll from the tirangle.

[engetudouiti]

Yes about bone flip issue , I may miss use it. (but you need to set bone as rest pose)

Though you do not think it is important,

but the less deform , the more reliable bone, when user change from edit mode to pose mode (zero pose)

[Thomas Larsson]

Yes, the problem seems to be that pole target and elbowPoleP bones don’t remain at their rest positions. But that is not a new problem, my illustration above was with G3F and stable 1.6.1. The solution is probably to figure out the location and orientation of those bones after constraints, and then make that the rest pose. But not now.

[engetudouiti]

OK, I do not remember which version worked (or if it only made with beta of old versions, when you think how snap work with you customize MHX as daz manner. ( I remember after you decide to change mhx as 2 separate bones as same as daz way (devide as bend and twist) ,and how you cupselize >> edit pole target >> how snap worked.

I do not quick reply about this, but it seems one problem you may need to consider .

I hope new way can work stable and many MHX user can approve current way.

(basically, pose rig and deform rig may need to use same bone roll, but new way is not, so it may cause un-expected deform I still think so, but it seems only way to use DAZ jcm correctly)

‌

[Alessandro Padovani]

note. Just a side note. Personally I use pole targets parented to the master bone, because I want full control. I don’t like poles moving around if it’s not me keyframing. Also, it’s perfectly normal that there may be bad deformations if the poles are not positioned correctly, that’s what poles are for.

[Thomas Larsson]

The snapping problem seems to be solved now. The elbowPoleP and kneePoleP bones have been moved to the right position in edit mode. The forearm twist was a separate problem which should also be solved.