Parent Influence Bug - rc.3

[nuxttux]

Hi - I recently noticed an odd behavior with the rotation influence of the parent effect.

Create two objects, exp: rectangle + null. Add parent effect to the rectangle, set Rotation influence to zero then select the Null as the target. When you rotate the Null, the rectangle will rotate off axis - so it doesn't seem rotate in relation to the Null, but it still rotates.

Likewise, if I set the position x,y influence to zero, set target to Null then rotate the Null, the rotation is off.
I think it rotate from the upper left corner in this scenario.

[rodlie]

@pgilfernandez are you able to look into this?

[pgilfernandez]

@pgilfernandez are you able to look into this?

👍🏻 I'll have a look at it this weekend

[pgilfernandez]

It's done! I'm pretty happy with the result, have a look at the PR #710

@nuxttux if you test this new implementation with a previously created project, bare in mind that you may need to reposition your children (the objects with the "parent effect" applied) as I changed the maths behind the effect and how "they evolve" when dragged, repositioned, etc...

Let us know!

[nuxttux]

The Linux build is still ongoing, I'll test it once it's done. Looking at the demo however, it seems to be behaving as expected.

[nuxttux]

All right - I tested it and it works as expected, for the most part. Something else did change.

Previously: If you parent an object to a Null, place that Null inside a group, you can move the group and it will not affect the Null-Object relation.

Currently: Same scenario, except that whatever you do to the group will reflect on the Null-Object relation.

See example.

[pgilfernandez]

All right - I tested it and it works as expected, for the most part. Something else did change.

Nice

Previously: If you parent an object to a Null, place that Null inside a group, you can move the group and it will not affect the Null-Object relation.

Currently: Same scenario, except that whatever you do to the group will reflect on the Null-Object relation.

See example.

Well, yes, this change was done in purpose to follow a more robust parenting system. BTW I understand it may confuse or be a too big change for animators... at this point I think I'm not in position to make a decision about it, maybe @rodlie has a clear opinion about it. Anyway, I understand your workflow and I was able to keep that feature while "fixing" all the previous odd behaviors.

If any other one wants to test this last proposal, I made a simple project to test it out parent_influences_v04.friction.zip
@nuxttux , could you confirm this is what you want and like?

I've added a new commit with this "fix", test it and let us know if it works now perfectly under your perspective and let @rodlie decide which to keep.

[nuxttux]

could you confirm this is what you want and like?

Yes, that is the behavior that I prefer - for what I do.

The "Follow Object" effect already behaves like the more robust parenting system - as in, all transforms applied to the null's group are still reflected in the Null-Object relation.

[rodlie]

So, we are ok on how it works in PR #710 ? if so I will review the code this weekend.