I Just thought a cheap way to implement it.

1. Select a point.

2. Delete it.

3. Insert the point back.

4. reconnect the point with previous splines

In fact, you can do it manually by now.

22 posts • Page **2** of **2** • 1, **2**

I Just thought a cheap way to implement it.

1. Select a point.

2. Delete it.

3. Insert the point back.

4. reconnect the point with previous splines

In fact, you can do it manually by now.

nine said that he didn't want to spend hours (potentially) manually tweaking his model

"We're so sorry, Uncle Albert,

But we haven't done a bloody thing all day."

--- Paul McCartney

But we haven't done a bloody thing all day."

--- Paul McCartney

- pndragon
**Posts:**591**Joined:**Sun Dec 05, 2004 1:27 am**Location:**North Carolina

Ok, let's put that straight

I'm not sure what flash does, but it either removes controlpoints (by doing something like degree reduction) or moves controlpoints (maybe by doing something like degree reduction followed by something like degree elevation). As far as a single spline is concerned, there's no difference between 2D and 3D - in both cases the spline is something 1-dimensional.

What makes a difference is that surfaces are 2-dimensional. They are defined by a "mesh" of splines. You can remove a point from a spline and the spline will still be connected, but you can't do that easily with a mesh (you'd get a hole in the mesh).

Maybe saying that working with surfaces is much more complicated than working with curves is more correct than saying 3D is more complicated than 2D.

I can think of 3 possibilities to perform that "smooth" operation for meshes:

1) Remove points:

If the mesh was made only of rectangular patches, it would be easy to remove a controlpoint and connect the other splines with hooks - this would make the surface smoother. But what if there are 3- and 5-point patches or hooks. I think it would be too difficult to cover all "special cases" to do that automatically (I am talking about removing a single controlpoint!), but it is quite easy to do it manually.

2) Move points:

In a regular mesh, each controlpoint is part of at least two crossing splines. What if one spline would become smoother by moving the point upwards and the other spline would become smoother by moving the point downwards?

3) Adjusting the tangents:

Yep, this would work. Probably there is an analytical solution to get the smoothest surface, but this is beyond my math skills. If someone has the necessary understanding of the math involved, please let me know!

All I can do is trial and error - and it took quite a long time until I came up with the current algorithm (that computes the tangents right now). I am sure that it can be improved and I am willing to do more experiments in this area, but for now I am quite happy with it as it is and there are more important things I have to work on.

I really don't want to discuss this subject and longer, but if someone likes to implement that "smooth button" I won't stop him.

I'm not sure what flash does, but it either removes controlpoints (by doing something like degree reduction) or moves controlpoints (maybe by doing something like degree reduction followed by something like degree elevation). As far as a single spline is concerned, there's no difference between 2D and 3D - in both cases the spline is something 1-dimensional.

What makes a difference is that surfaces are 2-dimensional. They are defined by a "mesh" of splines. You can remove a point from a spline and the spline will still be connected, but you can't do that easily with a mesh (you'd get a hole in the mesh).

Maybe saying that working with surfaces is much more complicated than working with curves is more correct than saying 3D is more complicated than 2D.

I can think of 3 possibilities to perform that "smooth" operation for meshes:

1) Remove points:

If the mesh was made only of rectangular patches, it would be easy to remove a controlpoint and connect the other splines with hooks - this would make the surface smoother. But what if there are 3- and 5-point patches or hooks. I think it would be too difficult to cover all "special cases" to do that automatically (I am talking about removing a single controlpoint!), but it is quite easy to do it manually.

2) Move points:

In a regular mesh, each controlpoint is part of at least two crossing splines. What if one spline would become smoother by moving the point upwards and the other spline would become smoother by moving the point downwards?

3) Adjusting the tangents:

Yep, this would work. Probably there is an analytical solution to get the smoothest surface, but this is beyond my math skills. If someone has the necessary understanding of the math involved, please let me know!

All I can do is trial and error - and it took quite a long time until I came up with the current algorithm (that computes the tangents right now). I am sure that it can be improved and I am willing to do more experiments in this area, but for now I am quite happy with it as it is and there are more important things I have to work on.

I really don't want to discuss this subject and longer, but if someone likes to implement that "smooth button" I won't stop him.

- sascha
- Site Admin
**Posts:**2792**Joined:**Thu May 20, 2004 9:16 am**Location:**Austria

I hope you can forgive me for perhaps pushing "the smooth button" a little to much. Perhaps I need a Smooth button myself or/and we need one all of us. For me it is easy to get exaggerated from new ideas - but really not so new - I have been thinking of this for perhaps a year - but not discussing it with someone(s)

1. Select a point.

2. Delete it.

3. Insert the point back.

4. reconnect the point with previous splines

Will this really make it?

Here you can have a look on a test:

http://www.jpatch.0nyx.com/smooth_test.htm

And I am content with Saschas replies - he thinking loud about it - I do not want to push him in this. He knows best what is to be done just now to get JPatch beccoming better and better.

1. Select a point.

2. Delete it.

3. Insert the point back.

4. reconnect the point with previous splines

Will this really make it?

Here you can have a look on a test:

http://www.jpatch.0nyx.com/smooth_test.htm

And I am content with Saschas replies - he thinking loud about it - I do not want to push him in this. He knows best what is to be done just now to get JPatch beccoming better and better.

- nine
**Posts:**133**Joined:**Mon Jul 04, 2005 2:28 pm

22 posts • Page **2** of **2** • 1, **2**

Users browsing this forum: No registered users and 1 guest