Plaster pattern on a barrel vaulted ceiling

[Erich]

Does anyone have a good method for modeling a plaster pattern on an elliptical vaulted ceiling?

I considered using a complex profile and a curved wall but the profile would merely be extruded along the curve not normal to it.

I could do it with GDL, but since there isn't a command to allow for a bprism with multiple radii, the maths would get a bit harry.

Thoughts?

[Dwight]

I suggest:

Make the vault with a complex profile wall.

Make the detailing pattern as a slab that intersects with the entire arch. [deep enough to extend the vertical distance of the arch]

Create a second, inner arch as a complex profile wall that is exactly the plaster pattern trim depth inside the initial arch. [By reducing the perimeter of the initial arch definition fill.]

Solid Element Operation subtract the descending part of the plaster pattern slab with the inner arch.

Solid Element Operation subtract with upward from the outer arch.

[Dwight]

Enough, already, with the cheap and nasty!

Assuming those parts are painted wood, you could model each component as a tilted beam or column to match each local slope. That's not so difficult since it has quadrilateral symmetry.

[Erich]

Dwight,

The cheap and nasty doesn't quite work for me. For it to be accurate, I would need to distort the pattern to match the plan projection at the arch bleah! .

Modeling piece by piece is an option, and perhaps one I will need to undertake. However, since the pattern is not yet fully settled, I am not certain I want to spend the time only to have to redo it later.

Since the pattern I show is based on the vault flattened out, I am considering sectioning it, sticking it into GDL and rotating each section. The vault itself is already a GDL object and I can get individual section angles from the faceting of the curves in the building section. I was just hoping for something a bit easier.

If only we could bend slabs as easily as we can bend walls....

[Dwight]

There's only a dozen unique parts - it can't be that hard to model.

[Dwight]

Modelling Wizard Ralph is asleep right now but tomorrow morning he will show the OBJECTIVE way to do it.

A real solution.

[Anonymous]

If the results don't have to be dead on it looks like it could be done by using Dwight's method but using profiled beams (and perhaps columns in the central part) instead of slabs to take the cuts out of the inner arch.

A relatively crude version could be done with a single beam profile for the lower part and a column for the center. For better resolution it could be done in narrower strips. A side benefit is that the columns and beams can do the cuts automatically by adjusting the priority settings. Not a big deal but it saves doing the explicit SEOs.

Attached is a quick example just using a single beam profile repeated and adjusted (sort of) to match the slope of the arch.

[Ralph Wessel]

Dwight wrote:
Modelling Wizard Ralph is asleep right now but tomorrow morning he will show the OBJECTIVE way to do it.
A real solution.

Thanks Dwight. Yes, OBJECTiVE provides an easy way to solve this. It does pretty much what you wanted, i.e. bending a slab to a curve. If you draw the shape you want with a slab (or fill) and make an Elevation profile with OBJECTiVE's profiling tools, it can be bent to a curve (optionally preserving its proportions). Refer to the attached image.

However, OBJECTiVE can only bend the shape to a s single curve and your ceiling is elliptical, so the solution takes a bit more work. I'll illustrate this in the next message.

[Ralph Wessel]

Dwight wrote:
Modelling Wizard Ralph is asleep right now but tomorrow morning he will show the OBJECTIVE way to do it.

To fit the profile to an elliptical ceiling using OBJECTiVE, do the following (referring to the attached image):

• 1. Draw the shape of the ceiling laid out flat (as you have already done) with a slab or fill.
  
  2. The ceiling shape will have to be broken down into arcs (which approximate very closely to the elliptical shapes). I created a polyline of arc segments, then split the roof shape to the lengths of each arc in the segment.
  
  3. Next make the fills/slabs into profiled objects. For each segment in turn, select it and click the menu OBJECTiVE > Component > New Profile. In the New Profile dialog set the profile type Elevation, enter a profile name, pick the library to store it in (Embedded is convenient), and a default thickness.
  
  4. When you click OK, OBJECTiVE will prompt you to click two points indicating the bending axis (the direction of the plane the shape is bent in). In this case, it is straight across the ceiling profile. The new object is created after the second click.
  
  5. When the new profiled objects have been created, open their settings and check the option to Keep proportions so the shape is not distorted as it is bent to a curve.
  
  6. Now we need to tilt all the objects to an upright position (so we can look along their length). Select all the profiled objects and click the menu OBJECTiVE > Tools > Rotate. The rotation is much like ArchiCAD's 2D rotate - click the two points to the axis of rotation, then a third to set the rotation angle (you can enter the angle in Tracker too).
  
  7. Now we take each object in turn and align it to its equivalent curve in the ceiling polyline. Move the object origin to the arc origin and rotate it so it passes through the arc end.
  
  8. Then click the stretchable node on the profile and use it to bend to the arc curve.
  
  9. The finished segments should look like this in 3D.
  
  10. Then use OBJECTiVE > Tools > Rotate to tilt the objects back to the level plane again.
  
  11. The finished ceiling in 3D.

[Erich]

Thanks all

I knew Objective would be able to deal with this, unfortunately the budget doesn't allow for it right now (sad isn't it). However, Ralph has given me a reasonable GDL solution. I can use a series of local curves and then use a series of bprisms set to these local curves. This will result in a minimal number of bprisms and a reasonably distortion free result. Revisions to the pattern would be fairly simple to accomplish merely requiring updating the coordinates for the bprism.

Mathew, I had not considered using profiled beams and Dwights method. That is a good thought and one I will keep in mind.