How can I get rid of sub control points on Form?

[aiamlo]

Hi,

A while ago Fusion360 introduced some kind of sub vertices control points on FORM, which I don't have a clue about, sometimes those get generated in certain faces which increases density but cannot be modified unless you use the straighten, flatten, and other tools, edit form doesn't work:


This is the box display:

And the blue line is how the model should look if it wasn't for the annoying sub-control points:

I've used Alias Speedform for a few years, and never got this kind of issue, and is pretty annoying to see this in F360 because none of the "utilities" tools like repair or making uniforms get rid of those control points, and I have to export the control points as .obj quads and import it again to clean the mesh every time I get those, which is very inconvenient.

I hope someone can help me and if so thank you very much in advance because I haven't seen anyone posting about this problem.

[jeff_strater]

I am not aware of any change that would cause such things to appear.  Can you share an example model here, and also give some info about what TSplines operations you did?  (e.g. Subdivide, Insert Edge, etc).  Thanks.

[glenn-chun · Accepted solution]

@aiamlo The sub control points you're referring to are called tangent handles.

• When they are linked (aka aligned or constrained), they are displayed as small green circles in Fusion. Their positions are automatically updated to maintain the tangent directions when the surface is modified.
• When they are unlinked (aka unaligned or unconstrained), they are displayed as small red squares in Fusion. Their positions are NOT updated when the surface is modified.

When a T-Spline surface is created, all tangent handles are linked by default in Fusion. If you manually move a linked tangent handle, it will become unlinked.

Looks like some tangent handles have been unlinked in your model. Here's how to link them again:

1. Invoke the Edit Form command.
2. Set Selection Filter to Vertex (or simply press Alt+A on Windows or Control+A on Mac).
3. Window-select all vertices, and you will see Link tangent handle appear on the Edit Form dialog.
4. Press the link icon on the dialog, and all tangent handles will be linked.

Glenn

[adam.helps · Accepted solution]

These "sub control points" are called tangency handles. You're correct that SpeedForm didn't have them, because SpeedForm only uses subdivision surfaces; only T-Splines are capable of having tangency handles.

I'm sorry that the tangency handles are bothering you. They're not a new feature; they've been around quite a while. They do have a purpose: when we disable tangency handles, all creases (and therefore all open edges) are "perfectly flat" (no curvature) as the surface approaches the crease, which makes it impossible to do things like perfect arcs or spheres.

Often, a user doesn't care about controlling boundary curvature, and would be happy to get rid of these extra control points. To support this, we allow you to mark a tangency handle as "don't care," which we call "linking" it. Tangency handles in green are linked, while tangency handles in red are unlinked. The reason you have some curvature in your surface is because you have unlinked tangency handles.

To link them, use 'Edit Form' and change selection to vertex mode, which makes the handles appear. Select all the tangency handles which are causing you problems. You'll see a "Link Tangency Handle" icon appear. Click on that, and they'll turn green again, and your surface will go back to flat.

Note that some tools, like "match," might cause tangency handles to become unlinked. This is necessary; if the surface you're matching to has some curvature, it's impossible to get a good match with linked tangency handles.

I hope this helps.

[aiamlo]

---

@adam.helps wrote:

These "sub control points" are called tangency handles. You're correct that SpeedForm didn't have them, because SpeedForm only uses subdivision surfaces; only T-Splines are capable of having tangency handles.

I'm sorry that the tangency handles are bothering you. They're not a new feature; they've been around quite a while. They do have a purpose: when we disable tangency handles, all creases (and therefore all open edges) are "perfectly flat" (no curvature) as the surface approaches the crease, which makes it impossible to do things like perfect arcs or spheres.

Often, a user doesn't care about controlling boundary curvature, and would be happy to get rid of these extra control points. To support this, we allow you to mark a tangency handle as "don't care," which we call "linking" it. Tangency handles in green are linked, while tangency handles in red are unlinked. The reason you have some curvature in your surface is because you have unlinked tangency handles.

To link them, use 'Edit Form' and change selection to vertex mode, which makes the handles appear. Select all the tangency handles which are causing you problems. You'll see a "Link Tangency Handle" icon appear. Click on that, and they'll turn green again, and your surface will go back to flat.

Note that some tools, like "match," might cause tangency handles to become unlinked. This is necessary; if the surface you're matching to has some curvature, it's impossible to get a good match with linked tangency handles.

I hope this helps.

---

Wow, thank you guys, seriously! And not only by letting me know how to link them but also how to use them!

Now that I know that can be manipulated, like holly molly, this is a game-changer for me, so many things that I couldn't do before or that I had to rely on mixing Form and Surface and that was so time-consuming, I'm mind blown seriously.

I'm just wondering another silly question, how do I add tangency handles to a vertex? 

Thanks in advance!

[adam.helps]

---

I'm just wondering another silly question, how do I add tangency handles to a vertex? 

Thanks in advance!

---

Crease an adjacent edge. If it's not too close to a star point, then you'll get tangency handles along with the crease. You can get rid of them by uncreasing edges (which is impossible on the surface boundary, so sometimes you just have to link them).

It's also possible to get tangency handles near star points, but it takes even more creasing. You have to crease to ALL the edges around the star point to at least two steps out (and you have to keep on creasing if there are other nearby star points). It's technically possible to get tangency handles on any edge, if you keep creasing enough edges.

[aiamlo]

---

@adam.helps wrote:

---

I'm just wondering another silly question, how do I add tangency handles to a vertex? 

Thanks in advance!

---

Crease an adjacent edge. If it's not too close to a star point, then you'll get tangency handles along with the crease. You can get rid of them by uncreasing edges (which is impossible on the surface boundary, so sometimes you just have to link them).

It's also possible to get tangency handles near star points, but it takes even more creasing. You have to crease to ALL the edges around the star point to at least two steps out (and you have to keep on creasing if there are other nearby star points). It's technically possible to get tangency handles on any edge, if you keep creasing enough edges.

---

Thank you very much @adam.helps , appreciated! 🙂

[mapperes]

"Their positions are automatically updated to maintain the tangent directions when the surface is modified."

How the position of tangency handles (TH) are updated? It is notable that some control points on the face infer on the position of the TH, but which and in what ratio? That is, when we make a TH linked, how its position is determined? Note: I know about tsm files and "grip" on TH and control points.

[adam.helps]

This is a little tricky to explain. One approach is to observe that the DeRose-style subd creasing for perfectly sharp, straight creases is equivalent to adding a triple knot to the surface and positioning the tangency handles in a particular way. That is how the tangency handle positions are derived. Note that this only works for straight creases; creases that pass through a star point, turn a corner, or form a “T” can’t be represented with tangency handles.

Another way to figure it out is to run B-spline knot insertion at the location of the crease until you get a triple knot. You’ll have three points in a row. Now, move the central point back to its original position (as it was before knot insertion). You should now have a crease that will be equivalent to a subd crease in that direction. The tangency handle positions are thus sort-of derived from knot insertion.

If you do the same thing in both parametric directions, you’ll end up with a 3-by-3 grid of grips. Again, move the central CV back to its original position. All the surrounding grips should be at their linked positions.

[mapperes]

Thanks for the answer.

Reading again your explanation, I have assumed that "location of the crease" means one of the two T-mesh vertex that is incident to a crease edge, right?
And "triple node" means multiplicity equal to 3, that is, the local knot value associated with that T-vertex will be inserted into the local knot vector twice, correct?
Also, the "parametric direction" is that perpendicular to the crease edge, right?

So, I would like to confirm, please, what are the values of the local knot vector into which the knot of the crease T-vertex will be inserted twice. Supposing the parametric direction is horizontal, is it formed from left to
right or right to left? (The positions of points in the two cases differs in your tests).
Also, moving the resulting "central point back to its original position" implies that the other two "points in a row" (the tangency handles) will be moved together, right? Or the positions of the THs are the ones determined
by the knot insertion itself?

Thanks in advance. Best.

[adam.helps]

Triple knots do refer to multiplicity three. Creases with tangency handles are actually vertex-centric; it’s theoretically possible to create a crease at only one vertex. Creasing an “edge” just means creasing the vertices at either end.

The knot vector in which you’re inserting knots is the one perpendicular to the edge, yes. Inserting knots is not an order-dependent operation; it should give the same result regardless of whether you add them “left-to-right” or “right-to-left.” If you are getting different results based on the order then there is probably a bug. As far as the specific values in the knot vector, these will be derived from the knot intervals on the edges in the surface; they are not constant.

When I refer to moving the central grip back to its original position, you should leave the tangency handles exactly where they were (they do not move). After doing so, the central grip, the tangency handle, and the next grip beyond that tangency handle should all be collinear. This will cause the curvature of the perpendicular isocurve as it approaches the crease to become flat, which is characteristic of subd-style creases.

[mapperes]

Thanks for the answer.

According to the attached image (by Fusion), we consider that the node circled in red is the vertex with crease. Considering the parametric direction indicated by the arrow, we make the vector of nodes in the local domain of the basis functions for that node, considering the weight (nodal interval) of the edges indicated in the image {1,1, ½,½}, so we obtain a knot vector {0,1,2,3/2,3}. By double inserting nodes in the coordinate corresponding to the red vertex (2), we were unable to reproduce the positions of the corresponding tangency handles that precede and follow this vertex, according to the theory. In the image (and confirmed in the .tsm), they are exactly on the third part of their respective edges, but we only reach this result if we consider a uniform knot vector like (0,1,2,3,4).

I take advantage of the same image to ask how it works in the case of border nodes. The tangency handle linked to the border node is also in the third part of the edge, however, if we consider the border node, what would the knot vector be like, and in which knot would the (single) insertion be made?

Note, our objective is purely research, using the models to simulate deformable solids.

Thanks in advance.