When weight normalize is on in the jointcaptureweight spreadsheet, and I try to set the weight of some points, Houdini has a strange way of normalizing.

When I try to assign a value of 1 to some points, after normalizing the value is never what I assign to the joint.

I don't now what Houdini is trying to do, but when a point is assigned to a single joint and I try to assign it 100% to a different joint by typing a value of 1 in the spreadsheet, it assigns a value of 0.5 to most points and some other value like 0.4 to other points on both joints.

I can't get Houdini to normalize as expected, which is a value of 1 to the joint I assign it to, and a value of 0 to the other joints.

Does anyone know some secret trickery to make Houdini normalize properly ?

Thanks.

I'm not sure but it's best to log a bug.

edward
I'm not sure but it's best to log a bug.

Thanks, I’ll probably do that.

It was logged as a bug in 20.5.380 end October last year, normalizing still not working with the spreadsheet in 20.5.455.

This seems like a fairly basic functionality for rigging to me...

Not sure it's a bug, or I'm misreading the issue. Normalizing in this context means that if you add all weights assigned to a point the sum is 1.0.

So if you assign a 1.0 weight for joint A on a point and it already had a 0.5 weight for joint B the result will be.

A=1.0/sum_of_all_weights
B=0.5/sum_of_all_weights

A=0.66
B=0.33

So the sum of the point weights is now 1.0.

Yes of course, but this is not about the calculation being correct, but about the tool working the way it’s supposed to, so it’s at least useable.

If it were a tool to randomly create numbers that add up to a value of 1, it would be perfect, but it’s not meant for that.

I suspect whoever developed this thought the same thing, but normalizing like this makes spreadsheet weight editing completely useless; you can’t assign the desired weight to a point.