MocapBiped1 Capture Weights (Huh?)

[Atom]

Hi All,

I am learning the rigging tools and I thought I understood how the capture bone weights tool works. I wanted to test my knowledge by loading up a MocapBiped1 character and examining the capture weights on the mesh to see how the professionals at SideFX manage their bone weighting.

But when I activate the tool and select the bone I can see that the bone does not have enough influence to even affect the mesh it is deforming. At this point I don't even understand how the mocapbiped is even moving? With the currently displayed weights there is not enough coverage to capture the skin.

Shouldn't the capture weights of this rig extend out to cover the mesh areas?

Currently they don't. They barely cover the size of the bone let alone the full arm or leg.

 

What actually binds the bone to the mesh?

Edited August 3, 2016 by Atom

[malexander]

The bone weights were likely painted on, instead of using the bone capture regions.

[michael]

a few things:

1 - you'll likely never get away with just using the capture regions or capture by proximity - an amount of hand painting weights will always be needed.

2 - I wasn't around when this model was created or used as the mocapbiped mesh - but I have had to work with it and it has a few issues, it's got some crazy scales for one thing...

3 - my bet is that the weights were established with regions or proximity and then the geometry was moved in and out, back and forth around a bunch of different file and finally added to the mocapbiped, meanwhile the mocapbiped bones likely experienced some adjustments along the way...

so what you are seeing isn't really representative of what this kind of asset should look like under the hood, at least for standard capturing workflows.

[MrScienceOfficer]

3 hours ago, Atom said:

What actually binds the bone to the mesh?

The capture weights are stored as attributes(boneCapture) in the geometry itself(read in from the .bgeo file).  The deform node doesn't look at the capture regions, it simply reads the attributes from the geometry.

[Atom]

Ok, so does that mean the capture regions are fallback if the attributes do not exist?

[Guest tar]

Capture regions are analogous to paint bucket fill tool. Good for grabbing large areas quickly, but detailed work needs retouching.

[michael]

if the attribute doesn't exist then the deform SOP will just error out, it will not 'look around' to find cregions - that's what the capture SOP does - create the attributes based on the cregion (or proximity)

once the attributes exist the points will be deformed based on the transform of the region - but it will ignore the size.

 

[Atom]

Thanks for the replies.

I am getting back to this task and I have successfully added a capture layer paint to my geo.

About Painting...?

I find the paint tool extremely slugish. Is there any setting I can change to improve performance?

Is there any way to just skip the brush tool and do a "Flood Fill" for each bone? For instance I can't paint the inside of the mouth on the head of my character using the brush.

How do I clear something when I make a mistake? If I middle click, the Reduce feature seems to have no effect.

Is there anyway to make the Capture Region of the Capture Layer Paint node auto populate when you click on that bone?

Edited August 6, 2016 by Atom

[michael]

I find that removing all groups and attributes helps speed things up (though this may not be needed in the current versions)

to flood fill - there is a button on the capture paint node called Apply to All (I think - don't have H at the moment) but I'd avoid that as it's not very smart - for areas like the mouthbag I make a point group and use that in the capture paint or capture override SOP. Sometimes I use the Visibility SOP to isolate parts of the mesh.

I don't use reduce or erase - I just paint to a different bone - back and forth till I get what I want.

to populate the tool - use ctrl+MMB to select a new region.