Finally after much fiddling around I'm getting close to finishing a new Twist command - after fixing up a few more loose ends it will be ready and I'll then put out a new v3 beta.

Twist will be a new deformation tool, going under Transform > Deform > Twist, next to the Flow command.

The way it works is you pick a twist axis (defined by 2 points) and then an angle. Your objects are twisted with 0 rotation at the start point of the axis, gradually increasing until rotating by the full angle that you specified at the top of the twist axis. This is how a helix curve is shaped so it basically helix-izes your objects, and in fact if you twist a line parallel to the twist axis that line will turn into a helix.

Some examples - starting with this shape:



With a vertical twist axis, twist with an angle of 40 degrees gives this:



With 180 degrees:



With a whole bunch of degrees:



You can also use it to twist up something like this:





Here's an example of a more complex object being twisted:



Twist with a horizontal axis 180 degrees:



Twist along the corner-to-corner diagonal of 360 degrees:




Should open up a variety of new possibilities. I've just got a bit more work to do on the mouse interactive part where you can use the mouse to define the angle by kind of winding it around the axis.

- Michael

Ha that looks cool.... :-) Nice job Michael.

Peace,
Rich_Art. ;-)

Excellent work!

I was wondering how extruded parts could be twisted, as they don't show any more control points.

So do you manipulate the underlying NURBS equations to obtain a twist?

this look'S nice and useful.

Hi Brian,

> I was wondering how extruded parts could be twisted, as
> they don't show any more control points.

Yeah this doesn't work only by manipulating existing control points.


> So do you manipulate the underlying NURBS equations
> to obtain a twist?

Kind of - it's working at more of an analytic level but it's more of a surface fitting process - the deformer goes through a type of fitting process where it analyzes the deformation and fits a new surface to represent it, subdividing things down and adding more control points as needed until the new surface is within a close enough tolerance of the "ideal" procedural deformation.

One of the nice things about this process is that you don't have to worry about the control point structure of the input objects - even something made up of just simple 4 corner point planes will automatically get refined as needed to produce an accurate enough result.

There are quite a few different mechanisms in NURBS modeling that work in this kind of "iterative-refinement-until-accurate-enough" type of method - stuff like Sweeping, surface/surface intersections, curve projection, etc...

- Michael

Terrific! Bravo!

Hi Michael,

looks pretty impressive, I'm it will find many usages and users.

Thanks,
Felix

After Twist, that was Jerk, no? It's Rock'n Roll!

Step by step, I find again the functions that I liked in Amapi 3D.

It goes to miss some functions to create skins with a lot of irregularities; and why not, commands of erosion & corrosion (I don't know if it's possible with NURBS).

I would like always, a nice choice of bevels on letters (LOL!).
Have a good day.

This is awesome news Michael - I can't wait to try this out.

Just last night at the store I took a look at some of the new P&G fabric softener bottles:


http://www.packagingnews.co.uk/design/world-design-procter-gamble-fabric-softener/



http://popsop.com/2651

There seemed to be four or five brand names with the same bottle. I was wondering if this shape was formed by Sub-D or maybe by a hand-carved clay sculpture by industry artisans.

I'm not sure that a twist deformation of the nature that you are presenting was what formed this bottle, but I find it interesting that it gave me pause to consider last night.

Would you consider any sub-control elements such as a positional start and end to the twist deformations (perhaps defined by the inference on where the start and end twist-axis definition is made)?
And also, perhaps a logarithmic or varied control over the angular degree through the body of the twist - where an object set could start slowly on the twist, then accelerate to a more acute twist?

But none the less, this is great news! And it looks so smooth too. This will surely streamline a lot of procedural work in things like turbine blades and ornamental decoration.

Looks great Michael.
Any chance to have an easy in/easy out option ?

Hi Bard,

> It goes to miss some functions to create skins with a lot
> of irregularities; and why not, commands of erosion &
> corrosion (I don't know if it's possible with NURBS).

NURBS by themselves are not very well suited for that - mostly because a NURBS surface is made up of a rectangular grid of points, so to add a little divot in one area of a NURBS surface you actually have to add in an entire row or column of control points. Polygon or sub-d methods tends to be better for adding that type of detail because they can add more polygon faces in a small localized area.

Also corrsion or weathering is often added using displacement texture mapping or rendering shaders rather than explicit modeling of the weathering itself.

If you do want to model corrosion or little bumps and stuff like that, one of the brush-based sculpting programs like ZBrush or 3D-Coat are better suited for that kind of stuff rather than trying to do it with NURBS. NURBS modeling can work well for building the kind of ideal smooth starting surface though.

- Michael

Hi Mike,

> I was wondering if this shape was formed by Sub-D or
> maybe by a hand-carved clay sculpture by industry artisans.

Hard to say for sure, but I wouldn't be surprised if it was actually done with NURBS modeling anyway, despite it being something that's easier to do with Sub-d. That's just because for a pretty long time stuff across a wide range of manufacturing industries have been done with NURBS. Sub-d modeling in manufacturing is relatively new.


> Would you consider any sub-control elements such as a positional start
> and end to the twist deformations (perhaps defined by the inference on
> where the start and end twist-axis definition is made)?

At the moment it does stop at the bottom of the twist axis - anything below the bottom axis point does not get affected. But this is actually kind of problematic because the deformation mechanism doesn't really work that great with abrupt changes, it kind of messes up the fitting mechanism and causes a little hiccup right in that area.

So making the twist be limited to a certain zone would probably have to be combined with some kind of "ease-in/ease-out" method as well so there wasn't an abrupt transition from the area of no twist into the area of twist happening.


> And also, perhaps a logarithmic or varied control over
> the angular degree through the body of the twist - where
> an object set could start slowly on the twist, then accelerate
> to a more acute twist?

That might be possible, but it would be good to avoid controls for it that make you feel like you're taking a math quiz...

So far I haven't really thought about how to control something like that.

- Michael

Hi PaQ,

> Any chance to have an easy in/easy out option ?

Maybe - do you maybe have any good description available for what those In and Out parameters that you're showing there actually do?

Is that something related to some animation controls used in various other places in Modo?

- Michael

Thanks Michael,

> That might be possible, but it would be good to avoid controls for it that make you feel like you're taking a math quiz...

C'mon... couldn't you just imagine the suave look of a nifty scientific graphing calculator widget in MoI's side pane? ;-) I think that was a nightmare I had the other night.


I could imagine a check box in the dialog box like the one on the Spiral command, where if you check it you could then pick a twist angle at one end and then at the other.
As for an ease in / ease out control - If you could imagine a simple button in the dialog that says [with ease] that would add a spacing entry in the dialog, not unlike the one for the Fillet tool.
You could then ask for a numeric range by which to go from 0 to 88 mph in 6 seconds, so to speak, defining a soft transition on the ends - if the twist ends are inside an object's length.
For example: I could select an object, pick a start and end point, choose the rotation angle factor for one point and then the other, then specify the soft ease transition distance...
No, no... to much.... my ego would get drunk with power... ;-)

(simplicity)

Don't mind my creative rambling Michael, I'm still marveling at the twist tool's amazing potential.

Hi Mike,

> I could imagine a check box in the dialog box like
> the one on the Spiral command, where if you check it
> you could then pick a twist angle at one end and then
> at the other.

I don't think that will work though - in the spiral command you're picking 2 different radius values - in the twist command you don't pick any radius values at all, because the distance of the particular point being deformed away from the axis is actually used as the radius of the helix.

Since it is a deformation that is being applied to objects, it's a bit of a different thing than drawing a single curve like the spiral.

Right now I can't see how 2 different twist angle would work - I mean the way twist works is that at the base of the twist axis you get 0 rotation there and it goes up to the full angle at the top. To do "ease in/ease out" doesn't really mean to change the start or end angle so much as it would mean changing how the transition between them would work...

Controlling the transition speed is not really the same thing as specifying 2 different angles...

- Michael

I would imagine it as having the center axis pole (Like when creating a helix) with a point that slides up and down the pole (Like the loft seam point alignment points). The sliding point would define a percentage of the angle at the placed point.

Hi Burr,

> The sliding point would define a percentage of the angle
> at the placed point.

Not sure that would work, because that's would be some kind of linear step that would probably not provide for a smooth gradual start or end.

Something that does "ease-in/ease-out" probably needs to be based on more like some kind of cubic blend that would start with only a little bit of "angular velocity" at the ends, and do more near the middle.

- Michael

A "bulge slider" for twist..... (well, 2. One for in and one for out)

Hi Burr,

> A "bulge slider" for twist..... (well, 2. One for in and one for out)

Maybe - if I can figure out a some kind of setup that has a scale factor in it that can be applied to it...

But I'm not sure yet how that would work.

- Michael

Hi Burr,

> A "bulge slider" for twist..... (well, 2. One for
> in and one for out)

The other problem with this is that it would probably mean adding in an extra step to the entire Twist command flow just so you could possibly tweak this value.

Right now the way it goes is you enter in a twist angle and once you do that you're all finished...

It doesn't seem very good to add in an entire extra stage just for twist acceleration tweaking.


I think it will be possible to add an option though to limit the twist to only happen inside of the axis range and when you do that there will be an automatic gradual ease-in/ease-out built in, here's an example twisting just the upper half of this thing:



Or here with twisting only the middle area:



That will allow for things like this - limited range twist with a twist axis between these 2 points:





- Michael