[Bf-funboard] Some skinning feature ideas

[Benjamin Tolputt]

 > Really? Try painting weights in maya sometime.

I have, it sucks, I don't want it.

However, the ability to weight vertices completely to a single bone 
while weight painting I find invaluable. Let's take a workflow I use 
frequently (and I see in a lot of basic rigging tutorials), which is to 
get the application to make it's "best guess" at initial vertex weights 
and then edit them as necessary to make them closer to the desired 
deformations. This works well when you are simply adjusting weights up & 
down for individual bones, but becomes a pain (in Blender at least) when 
you need weight a series of vertices to only one bone. This (in Blender) 
requires you to go through all influences the vertex has and paint them 
all to zero. Whereas in Maya the workflow is to simply paint the weights 
for the bone in question to completely white (i.e. 1.0) which will 
automagically zero (or drastically minimize) the influence of other bones.

I too like the fact I do not need to work on normalization manually, but 
I don't think that need be a problem for implementing something that can 
fix this. Perhaps there could be a "auto-normalize" option/mode for the 
Weight Paint interface. When entering this mode/option, it will 
autonormalize the weights for you. Then, while painting, the 
normalization happens automagically. That is, when I see a vertex 
coloured completely red - I know the selected bone is the only (or nigh 
on only) bone influencing that vertex. If said vertex is coloured 
completely blue - I know that the bone in question is not affecting the 
vertex at all. Being able to see this while editing the weights would be 
an invaluable tool in my mind (and a major time saver).

The way I see it working is to have "adding" weight to the vertex 
influence do so, then normalize the rest of the vertex influences (i.e. 
weights them down) to match the "1.0" requirement. When vertex influence 
drops below a settable level (e.g. 0.001), it gets removed completely. 
When subtracting from the vertex influence, it normalizes the other 
influences (i.e. weights them up) to match the "1.0" requirement. The 
special case for this would be that when there is only the one influence 
affecting the vertex (and it is the one being subtracted), there is no 
normalization. In this case it is either 1.0 or 0.0 (as that is it's 
effective value anyway).

My thoughts on the matter are that just because Maya stuffed it up, 
doesn't mean the concept isn't sound. It might just be the case that 
Maya developers/designers are not perfect either, and as such borked the 
implementation.

Regards,
B.J.Tolputt