[Bf-committers] Revisions for the final 2.78: Rigify super_arm super_leg

[Lucio Rossi]

Hello,
Following the suggestions of Sergey I committed a new fix to the super_leg
and super_arm sub-rigs.
We got rid of the Child_of constraints altogether using copy_transforms on
an MCH-socket bone and parenting the ik ctrls to it.
This simplifies the code and solves both the problems we had:

- weird matrix multiplications
- funny interaction with Visual Transformations

you can find the fix here
https://developer.blender.org/rBAbf668cc3eef5

Same as before this is totally back-compatible. The user can try the new
limbs only choosing them explicitly.
Besides this fix doesn't mess up with rigs already made with super_arm or
super_leg limbs
Hope there's still time for porting...

Thank you :)
Lucio

2016-09-26 14:32 GMT+02:00 Sergey Sharybin <sergey.vfx at gmail.com>:

> Ok, have a more clear picture now.
>
> Let's port those changes over and call it all settled :)
>
> On Mon, Sep 26, 2016 at 12:37 PM, Lucio Rossi <lucio.rossi75 at gmail.com>
> wrote:
>
> > Hi,
> > 1. No, we don't apply the same child_of twice. It think it depends on the
> > interaction between the single constraint and the IK chain.
> >
> > 2. The fix "corrects" the ik-fk snapping functions in the case hand or
> foot
> > ctrl bones have the child_of constraints on top of it. Unfortunately we
> > very recently noticed that, independently of the modifications we made,
> > adding the two child_of constraints on top of the IK ctrl bones messes up
> > with the apply visual transform. This is not, I repaeat, dependent on the
> > fix or on the previous code, it's just the interaction with the
> constraints
> > (IK and Child_of) and the transform functions.
> >
> > 3. By not-default I mean that when the user chooses a Pitchipoy rig the
> > arms and legs are marked as super_limb (arm) and super_limb (leg) same as
> > before the commits. If, and only if, the user wants to try the new
> > functionality he/she has to give, esplicitly, the super_arm or super_leg
> > property to the those bones.
> >
> > To sum up everything works as before the commits unless anyone wants to
> try
> > the new sub-rigs...
> >
> > 2016-09-26 12:16 GMT+02:00 Sergey Sharybin <sergey.vfx at gmail.com>:
> >
> > > Hi,
> > >
> > > On Mon, Sep 26, 2016 at 11:46 AM, Lucio Rossi <lucio.rossi75 at gmail.com
> >
> > > wrote:
> > >
> > > > About the double matrix multiplication IMO it depends on the fact
> that
> > > two
> > > >
> > > extra child_of constraints have been added to the IK controller bones.
> > > > Though it works together with the fk/ik snaps we're experiencing some
> > > other
> > > > issue in the apply visual transform we're trying to address.
> > > >
> > >
> > > That would be weird if same child_of matrix is applied twice. If it's
> two
> > > different child_of, then multiplying matrix by itself is not a correct
> > way
> > > to go to a proper space.
> > >
> > > You mention some other transform issues, so the fix is not even
> complete?
> > >
> > >
> > > > Finally, please consider that the two subrigs introduced are not (up
> to
> > > > now) default for arms and legs in the Pitchipoy rigs.
> > > >
> > >
> > > We've passed two release candidates and few hours from calling a final
> > > release. We have to be strict, meaning: the fix should either be fixing
> > > regression or be dead-simple for inclusion. Currently this doesn't seem
> > to
> > > be the case here (especially assuming even with the fixes applied there
> > are
> > > still other issues).
> > >
> > > Did i understand correctly, that the fixes are in the area which is now
> > > became default? In this case, how much unusable things are prior to the
> > > fix? How much it's unusable after the fixes (you mentioned some issues
> > > still)?
> > >
> > > 2016-09-26 10:46 GMT+02:00 Sergey Sharybin <sergey.vfx at gmail.com>:
> > > >
> > > > > Hi,
> > > > >
> > > > > First two commtis are quite some code which does not address
> > regression
> > > > or
> > > > > trivial fix. So it is rather really risky to accept it now.
> > > > >
> > > > > The code itself is also weird:
> > > > >
> > > > > - You can clearly simplify the layer mask creation, and don't have
> it
> > > > > inlined as a list of False, False, False.... in multiple places
> > > > > - Multiple subsequent try/catch could be replaced with a single
> check
> > > > > whether `rigify_parameters` is in pbone. Using exceptions for that
> is
> > > > > totally wrong!
> > > > > - The goal is to have code pep8-compliant, and the new code is
> > clearly
> > > > not
> > > > >
> > > > > Remaining two commits is clearly less code, but they are still
> > > > suspicious.
> > > > > Never heard of getting square of a transform matrix (mat*mat), and
> > not
> > > > sure
> > > > > code was tested well enough after the change was done by all the
> > users.
> > > > >
> > > > > On Sat, Sep 24, 2016 at 3:16 PM, Lucio Rossi <
> > lucio.rossi75 at gmail.com>
> > > > > wrote:
> > > > >
> > > > > > Hi,
> > > > > > Adding to the last mail two commits that were missing...
> > > > > >
> > > > > > If it's possible include the following commits in the 2.78
> release.
> > > > > > They add a new super_arm and super_leg rigs with switchable
> > child_of
> > > > > > parenting:
> > > > > >
> > > > > > rBAff11927e8e07a46c8199d9681af639be52b6b1af
> > > > > >
> > > > > >
> > > > > > rBAfecd5c9f566e0075e91d6d5830ccc0a622932736
> > > > > >
> > > > > > And as said in the previous mail:
> > > > > >
> > > > > > rBA816df6cfe1054acffad2ec23edd63ae504ab325f
> > > > > >
> > > > > > rBA4ed120fafb157aeeb9fbdb072a0e75375d658e24
> > > > > >
> > > > > > Lucio
> > > > > > _______________________________________________
> > > > > > Bf-committers mailing list
> > > > > > Bf-committers at blender.org
> > > > > > https://lists.blender.org/mailman/listinfo/bf-committers
> > > > > >
> > > > >
> > > > >
> > > > >
> > > > > --
> > > > > With best regards, Sergey Sharybin
> > > > > _______________________________________________
> > > > > Bf-committers mailing list
> > > > > Bf-committers at blender.org
> > > > > https://lists.blender.org/mailman/listinfo/bf-committers
> > > > >
> > > > _______________________________________________
> > > > Bf-committers mailing list
> > > > Bf-committers at blender.org
> > > > https://lists.blender.org/mailman/listinfo/bf-committers
> > > >
> > >
> > >
> > >
> > > --
> > > With best regards, Sergey Sharybin
> > > _______________________________________________
> > > Bf-committers mailing list
> > > Bf-committers at blender.org
> > > https://lists.blender.org/mailman/listinfo/bf-committers
> > >
> > _______________________________________________
> > Bf-committers mailing list
> > Bf-committers at blender.org
> > https://lists.blender.org/mailman/listinfo/bf-committers
> >
>
>
>
> --
> With best regards, Sergey Sharybin
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> https://lists.blender.org/mailman/listinfo/bf-committers
>