Pole Magnets

[MaDDoX]

You know what we need to make quadrangulation truly outstanding? Arbitrary placement of poles. Interface-wise those would be simple colored spheres that we would place on top of the original voxel model, and during quadrangulation would work as "topology magnets". This marker would flag to the system that this is a spot where it's okay to have extraordinary vertices (vertices with other than four edges), in practice attracting more - or less - than four edges towards it.

On top of that, each of these "pole magnets" would have a strength setting. Zero Strength would mean pole absolutely unwanted (ie. regular four edges crossing), strength = 1 would mean an optimal spot for an Y-pole (3-edges), a strength of 2 would mean ideal star-pole (5-edges) spot and a strength of 4 or more would mean extraodinary poles (6+ edges) accepted - probably useful at the center of very flat areas that would never be deformed in animation. For feedback purposes a simple color scheme could be used, strength=0 would be blue, strength = 1 would be green, strength = 2 orange and strength = 3+ would be red colored.

This way it'd be exceedingly easy to define the flow of the actual topology, you could for instance set four 'magpoles' at the four imaginary corners of a shoulder and an elbow and be sure that the quadrangulation process wouldn't generate those unwanted slanted (diamond-shaped) polygons. Just think at the power of such a feature! It would surely make even the denser critics of 3DCoat rethink their points of view

[Tony Nemo]

You know what we need to make quadrangulation truly outstanding? Arbitrary placement of poles. Interface-wise those would be simple colored spheres that we would place on top of the original voxel model, and during quadrangulation would work as "topology magnets". This marker would flag to the system that this is a spot where it's okay to have extraordinary vertices (vertices with other than four edges), in practice attracting more - or less - than four edges towards it.

On top of that, each of these "pole magnets" would have a strength setting. Zero Strength would mean pole absolutely unwanted (ie. regular four edges crossing), strength = 1 would mean an optimal spot for an Y-pole (3-edges), a strength of 2 would mean ideal star-pole (5-edges) spot and a strength of 4 or more would mean extraodinary poles (6+ edges) accepted - probably useful at the center of very flat areas that would never be deformed in animation. For feedback purposes a simple color scheme could be used, strength=0 would be blue, strength = 1 would be green, strength = 2 orange and strength = 3+ would be red colored.

This way it'd be exceedingly easy to define the flow of the actual topology, you could for instance set four 'magpoles' at the four imaginary corners of a shoulder and an elbow and be sure that the quadrangulation process wouldn't generate those unwanted slanted (diamond-shaped) polygons. Just think at the power of such a feature! It would surely make even the denser critics of 3DCoat rethink their points of view

I was prepared to come down on the side of "No more new tools, perfect the engine" in the forum debate, but this looks a great idea! Retopology is one of 3DC's strongest assets and something that made it as controllable as that would have broad appeal.

[Ghostdog]

well I'll throw in my two cents and agree this sounds a real winner of an idea.

[Taros]

Maybe this thread could interest you...

-> http://www.3d-coat.com/forum/index.php?showtopic=3325

Andrew already commented it.

[GED]

yeah people have wanted something like this for a little while now.

[ShnitzelKiller]

That was over a month ago, any other news on this? That sounds amazing!

I'm all for ways to control quadrangulation. Perhaps by drawing some closed shapes in the general areas you want loops to be or something, and letting it work from that. That way quadrangulation of sharp edges won't result in that jagged diagonal quad problem.

[MaDDoX]

Thanks guys. Bump?

Maybe this thread could interest you...

-> http://www.3d-coat.com/forum/index.php?showtopic=3325

Andrew already commented it.

Yeah Taros I did read that thread before posting my own, the thing is that IMO quadrangulation control is actually more important than a superior quadrangulation scheme. Think about it, even with that improved automatic remeshing tool, for actual animation (which's where retopo really is needed) you'd still need to have better control of loops and such to get the best deformations possible.

My point is that the current quadrangulation algorithm is quite good already, its problem is that we can't control it and that wouldn't change with a better algorith. Asking for improved quadrangulation before better quadrangulation control features is completely counter-productive.

[Tony Nemo]

Thanks guys. Bump?

Yeah Taros I did read that thread before posting my own, the thing is that IMO quadrangulation control is actually more important than a superior quadrangulation scheme. Think about it, even with that improved automatic remeshing tool, for actual animation (which's where retopo really is needed) you'd still need to have better control of loops and such to get the best deformations possible.

My point is that the current quadrangulation algorithm is quite good already, its problem is that we can't control it and that wouldn't change with a better algorith. Asking for improved quadrangulation before better quadrangulation control features is completely counter-productive.

You can't control it when it's done but you can edit the results in in the Voxel room. My favorite joint area configuration of loops is for the loops being close on the outside of the bend (front of the knee, FI)and far apart on the inside (that moves the most). You can achieve a nice rounded knee or other 1D curve with this topology.