What a Mesh

Conversion of a mesh into faceted optimized in AutoCAD 2010

Hi,

I made another video supporting conversion from mesh into solid in AutoCAD 2010. One of the other possible options not yet covered is conversion to faceted optimized surfaces or solids. This conversion is pretty useful when we want to keep planar faces in the solid, and also provides the best roundtrip from mesh into solid and back into mesh, because the tessellation that AutoCAD has to create to convert the solid back into a mesh does not introduce much complexity.

I will elaborate on this concept in following posts. Until then, make sure you try this tip.

Experiment with non optimized conversions

In this example, I'm working in the design of a membrane for a stadium with AutoCAD 2010. My colleague and friend Brian Pene came up with a case that was as interesting as unexpected :-)

Based on it, I tried to make an axample where his discovery could be used.

Basically, when we explode a mesh, we get a collection of 3d faces. If these are converted into surfaces using Smooth not optimized, and then tessellated again, you can get an interesting pattern (as long as the faces are non planar). This pattern is caused by the way AutoCAD trieds to smooth the edges when reapplying the tessellation.

Check the example in this video.

Conversion from mesh into solid (part 1)

One of the most exciting options in the new Mesh modeling commands is the ability to convert a mesh into a solid or a surface. It is really a differentiator, since we can handle different topologies, and have workflows among them.

Once the mesh is converted, it can be manipulated with the Autodesk Shape Manager (ASM), which means that Boolean operations (Union, Subtract, and Intersect) and other operations for solids can be performed.

There are four options for converting to either solid or surface, which are smooth optimized, smooth not optimized, faceted optimized and faceted not optimized. Why so much choice? As a user, do I need to care? In fact, you do, if you want to take advantage of the whole power of this operation.

Let’s start with Convert to Solid.

In order to convert into a solid, we will use the command CONVTOSOLID, or access it from the Ribbon (it is located in the Mesh Modeling tab, in the Convert Mesh panel.

In order to be successful in the conversion into a solid, the object must comply with a couple of conditions. It has to be a watertight structure, which means that there should be no gaps in the model. If there’s a gap, then the operation will not be successful. The example below shows a mesh with a gap. It was done with a revolution, and as you can see, it is not closed. This is the kind of mesh that won't get converted into a solid.

 

The object can’t self intersect. Self intersections may happen really easily with all the flexibility and power for manipulation that our tools provide. The whole point is that you may not care, if you are going to remain as a mesh. We have already explained how mesh modeling is much more pervasive in the media and entertainment side, and in those cases, fabrication is not an issue. But when we try to create a solid, self intersection is not acceptable.

This is a clear example of self intersecting mesh.

The first option that appears in the right side of the panel will be by default Smooth Optimized. We’ll focus on this option today.

Smooth Optimized will create a smooth solid, and AutoCAD will attempt to stitch faces with G2 continuity. That ensures a solid with the least amount of Nurbs patches. This is especially useful when exporting into Revit. But we’ll leave that for other posts.

 I will show a couple of examples that may help understand how much we can optimize the amount of Nurbs patches. It will also help to keep in mind how to proceed when doing mesh modeling. If the final destination is Revit, we will want a low number of patches. If we will remain in AutoCAD, you don’t need to care!

In this first example, I started with a box, increased the smoothness level to 2, and did some manipulation of edges and faces.

When the model is converted into solid, it has 6 patches, which is the least amount of patches that a box can have.

When a crease is applied on the edges of an object like in the example below, you will notice that the solid has one more patch, in order to solve the creasing. Since creasing is basically affecting the continuity, it seems quite obvious that there will be an effect on the way AutoCAD converts into a solid.

Once we convert into solid, it is still worth noting how the conversion still provides a low amount of patches in the rest of the solid.

In the next example, I creased all the faces at the bottom of the box.

This is a pretty obvious case when modeling a building that may normally need to be firmly based on the ground. Something similar happens with the patches in the solid. In the previous examples, more patches were created next to the creasing. The same thing happens here. You’ll notice a second row of patches, simple because of the fact that the creasing affected all the faces at the bottom, thus having an effect over the whole ring of edges.

 

Another case to take into consideration is when you refine the mesh. If the refinement is done for the complete model, it will behave pretty much in a similar way to the first case.

This is a basic mesh primitive (box) that was refined when in level 2. 

 

The solid created with Smooth Optimized is exactly the same as if we had used the primitive in Level 2 with no refinement. 

AutoCAD can really make a good optimization of the faces in the mesh. You can see this in the example below.

In this case, we'll refine the box and make some manipulaitons that add more complexity to the overall shape.

After refining and manipulating faces, we still have 6 patches!

However, if we refine a specific part of the model, the optimization will be less predictable. Refining selection is in my opinion a very interesting way for working with less faces, and only adding complexity where needed. As I said before, you may or may not care about the amount of Nurbs patches in the object.

This is the mesh with a couple of faces refined.

 

The resulting solid has much more patches.

There’s a lot more to cover, but we’ll keep it simple now, and elaborate in following postings.

Mesh split and mesh extrude

This new video is about two commands that enable you to split and extrude faces in a mesh.

As usual, there's an HD version if you click on the link in YouTube

Creation of a mesh with an Edge Surface

Hi,

Today I'll post a video about the creation of a tensile structure using the Edge Surface. This surface was migrated into the new mesh topology, so we'll explore some of its new potentials through an example.

You can see the video with the following link:

If you go to YouTube, you'll be able to see the HD version.

We'll keep discussing the Edge Surface in following postings, so stay tuned.

Creasing a Mesh (part1)

Hi,

I’ve been quite active posting videos, so I now want to catch up writing some posts about the videos and some other interesting topics for any 3D user.

I covered creasing a couple of videos ago, so I’d like to write about it, and dig more into some details for Mesh in general. We already talked about how the fact of having a smooth, continuous surface is achieved by producing an approximation with polygonal faces by adding vertices and subdividing those faces. Continuity is definitely the most interesting characteristic of this scheme.

 

Strange as it may seem, both the cube and the sphere in the image above are topologically the same. The sphere started as a cube that was smoothened into Level 4. The whole structure of the object has not changed after these operations. That’s why we can go back into any level of Smoothness if needed.

Bus as we saw in the example in the video, there are times in which we may need to have certain edges or faces which don’t get smoothed. That’s when we apply Creasing. This command specifies that next to a selected edge, the subdivision needs to be done sharply, by adding steps, which can go from 0.1 into infinity. Actually AutoCAD can use a range from 0.1 to 2 for what we will call variable creasing. The more steps used, the more creased the surface will be. A Crease value 0 means that the object is smooth.

The command MESHCREASE can be found in the Ribbon, with the command MESHUNCREASE, which obviously removes the creasing from a selection. Both commands work either by selecting the geometry and then applying the command, or by selecting geometry after applying the command. Though it may be debatable, I normally use the command after having selected geometry. It may also happen that after the manipulation of a face or edge, we want to crease the selection, and since we already have it done, so it’s just a matter of clicking on the icon or typing the command.

Once MESHCREASE is invoked, we can define its value. The option Always means that we’re applying infinite creasing, meaning, as we said before, that the edges or faces selected are sharp. We can also input any value between 0.1 and 2.

If we need to change the value of any creasing applied, it can be edited through the properties palette, which will display a new category called Crease. Selecting the Type By level, we’ll see a new row called Level appear. That’s where we will input a value for variable creasing.

 

It is important to understand that creasing a Subobject of the Mesh will change the properties of the subdivision, and depending on how we make the model, we may have some challenges when trying to convert into a solid or surface. But it’s all about knowing best practices, so stay tuned for the next post.

Let's talk about Mesh (part 2)

During many manipulations of a mesh, you may want to reproduce a certain behavior along an axis, which means that you will need a solution for axial or central symmetry. Although AutoCAD doesn’t have a specific tool for editing at both sides of an axis, there are certain tricks that will do the job for many cases.

The most obvious case is when we need to move faces on the same direction. That’s pretty easy, since we just need to select these elements and move them.

However, other situations may get trickier, and that’s when we need to be a little smart and operate in a different way. If we want to move or enlarge objects in opposite locations, we can use the Scale Gizmo. The whole point here is how to locate the Gizmo to the center of the selection. We’ll just use snaps! Mid Between Two Points is maybe the newest snap (introduced many versions ago), and was just a blessing. Do you remember when we had to draw an auxiliary line and snap to its midpoint, and then delete it? Ouch!

Once you have relocated the Gizmo to the new position, the Scale option will do the job, if you move mainly along the axis that is somehow normal to the selections. You could also work with the whole plane in the Gizmo too, depending on what you want to achieve.

In this example, the selection of the Mid Between Two Points is done by clicking on the grips in both faces.

 

Then, just click and drag the axis, and see the changes in the model. Yeah... it's AutoCAD!

 

If you have more than a face selected on each side, you will need opposite points that when snapped with Mid Between Two Points, will deliver the real center of the whole selection. The image below shows one of the possible options.

 

This looks so good, that I may try to do something else. Why not select two more faces along a central symmetry, and apply the same process along the Z axis? Here are the results.

 

 

Have fun and see you in the next post!

Let's talk about Mesh (part 1)

Since the blog is called What a Mesh!, let’s start by mentioning a couple of things about one of the most interesting features in AutoCAD 2010, called Free-Form Design.

Many requests were coming to the Product Management team about the ability to work with smooth geometry. There could have been many potential approaches for this, and one of the decisive reasons to work with subdivision surfaces was to enable our users to work with a very intuitive approach to modeling. We have seen how 3D adoption has increased dramatically during the last years, and we definitely want to provide our users with tools that will make that transition easy. The fact that Free-Form Design enables you to sculpt mesh primitives with a variety of tools makes it become a very tactile approach to modeling. Sculpting with some sort of clay is one of the first exercises a kid does at school, due to the same reason. It’s intuitive, it unleashes the creative process, and it is fun!

Subdivision surfaces have been around for a while in the Media and Entertainment industry, since it’s a very cost efficient way to represent and render a smooth surface by using a series of planar faces. You can increase the level of smoothness by just a click, and get back to a previous level of smoothness if appropriate. In following posts, I’ll work with specific examples in order to introduce not only the features, but also best practices.

The image below shows a primitive (in this case, a box, to the left), that has a preconfigured amount of subdivisions. We will call this Level 0. The model in the middle is the same box with a Level 1 of smoothness, and the model to the right is the maximum level of smoothness allowed, Level 4.

 

If we refine the smoothed meshes (and we’ll talk about this later), we would get a radically bigger amount of faces in each object. For example, the Level 0 had 64 faces, the Level 1 had Level 0x4=256 faces, and Level 4… guess… Level 0x4x4x4x4= 16384 faces!

 

As you can see, the number of faces can quickly become quite impressive, and, by the way, I really don’t believe you’ll ever need so much level of detail. Best practices are coming in following posts. Stay tuned!