Yes Remus, definitely a good idea assuming perfect geometry. Wo3Dan, your idea about building the core in wedges is sound. Could the junctions between the wedges be hidden?

One of the main problems here is that I am building the model from 2d drawings which I thought were exact but which, it turned out, were not. I just imported profiles from the 2d drawings into SU. So now I'm having to go through the whole model (what's posted here is just a fraction of the building) correcting geometry.

I've replaced the original model on this thread, with a new (hopefully) corrected version.

Wasn't sure what you meant about
@unknownuser said:

slanted in two directions
. Maybe it's corrected now.

And Tom, I like your idea of getting the groups out of the model and working on them at a distance. What do you mean when you say "open one group"? Do you mean explode?

Maybe there is no "easy" way. At least there are options

Thanks Remus, that method works in simple cases. Here is a portion of the model to give a better idea. Two of the cuts have not yet been made.[attachment=1:u1vn1rc0]<!-- ia1 -->[attachment=0]wall beam cut.jpg<!-- ia1 -->[/attachment:u1vn1rc0]

wall beam cut.skp

There have been similar discussions before.
Here is the problem.
Its a big model with lots of complex geometry.
Everything is carefully grouped and layered.
Lets be specific...a beam passes through a wall.
You want to cut the wall with the beam but both are grouped.
Exploding the groups is too dangerous.
Intersect-with-model option too dangerous.

My methods so far:
1 open the beam group for editing
copy relevant parts of the beam to clipboard
open the wall group and paste-in-place the relevant beam parts
with the beam parts and the wall selected, use intersect-with-selected
deleted unwanted beam parts and cut the hole in the wall
close group

2 select both groups and use intersect-with-selected
the intersection line is created outside either of the groups
with both groups closed, select the line of intersection
cut it to clipboard, open wall group and paste-in-place
cut the hole in the wall and close
(the cut faces must be drawn manually)

Someone must have a better method

I wasn't thinking you could improve your method. I was just wondering whether you know the various methods of selecting objects. See here:
http://download.sketchup.com/OnlineDoc/gsu6_win/gsuwin.html
I think your solution is great
If I think of any good problems I'll send them your way.

Thanks for a new trick Coen
with the rotation tool, pick the centre of rotation and then control the axis of rotation by holding down lmb and sliding in the required direction. Great! I've always rotated the model to get a different axis, which can be annoying.

Yes very nice solution Teezer, And very thorough. Also nice tut. The rolling process you used is accurate but I can see tedious. I guess you know about the two methods of box selection: from the left and from the right. That could speed things up. You may get a job at MIT

Nice Teezer! You have staying power
The chiral looks tasty. But you only gave us the briefest desciption of how you made it. Give us more details if you can face it!

Yes I like it

Yeah great, both those methods will work. The long-windedness is not too bad especially the "replace selected" option. Anyway it's the price we pay for not thinking ahead

Is there a way to make two or more components "not unique" ie, identical, after they have been created.

For example if you built an object, made it a group, multiple copied it, did some other stuff, then later decided you wanted to edit all instances of said object. You could make them all components, but could you make them all the same component?

I don't think you hijacked the thread...it's a team effort! The great thing is that there's always more than one way to solve any problem.

Yes, nice solution Teezer! And nice tut Simon. Teezer, please define "simple minimum solution". Is this a math term? A physics term?

It really makes sense to think of the hexagon vertices lying on a series of circles.

How about this refinement of your process.

1. Instead of drawing circles, draw polygons. In this case a polygon with 24 sides.
2. The "hand-stitching" process can be mechanized. Just draw two adjacent zigzags up the stack of polygons and multiple rotate/copy them around the perimeter(use rotate and control to get the + sign, click on the centre of the polygon which SU will find for you, start the rotation process, enter the relevant angle, then enter 6x)

Who would have thought a roll of chicken wire would be so absorbing

I couldn't help doing the math.

This method works because the distance from vertex to vertex across a hexagon equals twice the length of the side.
For a nanotube with N repeat units, we need a polygon with 3N sides.
The angle at the centre of this polygon is A = 360/3N
Using trigomometry,
So the point is that we can now build a nanotube with any number of repeat units, with a side of any length.

Of course the work is not over. What about the other types of nanotube? Armchair for example!

@unknownuser said:

I know this is not what you wanted to show us here. but I figured out a nice way to draw the form (chair leg?) you created in your tutorial in another way...

It's cool that SU automatically models the shape perfectly using the move tool
I tried the same using sandbox>from contours, but the result is messy

oops!

nanotube.skp

Inspired by Simon here's a shot at modelling a nanotube

Carbon nanotube - Wikipedia

(en.wikipedia.org)

It's not perfect - a bit of trial and error involved. This method is for the armchair variety. For the zigzag or chiral some tweaking would be needed.

Hey Pilou I can't believe chaoscope
The possibilities for beatiful pictures are endless!

And Simon, about the nanotube. I think it is worthy of a new topic! See Tutorials

Thanks Pilou
You are a source of great software! This one looks so good. Never before has there been so much free stuff!
Also I think your artwork is really great and unique I would say

I have just had a look at K3DSurf. It is really good. It is beautiful that such amazing shapes can be generated by mathematical functions!

Here is another bit of fun that you might enjoy. I thought that if the icosahedron was truncated, what is the object that is truncating it?

what truncated the icosahedron?

Topmod does look very interesting will definitely look into it.

Taff, there are a lot of interesting things here. Like when does SU decide that two points are close enough to become one point?

And about precision. Plato and Archimedes contemplated these perfect geometries. I think they might be disappointed that even now with all our technology, even in the virtual world of computer graphics, the only place a perfect shape exists is in still in our minds!

I would guess that even with the best possible software the position of any point is just an approximation to some pre-assigned level of precision.

I don't know how SU works...presumably a circle is represented within the program by its equation. Then the intersection of two circles is found by solving their equations simultaneously. AFAIK all programs use methods of successive approximation to solve equations. Which means that the point which is calculated is only an approximation! And when it comes to representing these things on a screen, the pixel is the limit. So those nanobots may have to wait before they can get perfectly shaped soccer balls!

Enough philosophy. I tried the same construction in another application...Moi, which btw is brilliant, but quite different from SU. And the precision issue didn't even come up. It just found those circle intersections straight off and rotated those hexagons exactly to the point without any doubt. One of the things I love about 3D work is that all applications are different and have different ways of doing the same job

Cheers,

Steve

Thanks Taff, your constuction method is great...I didn't know about the golden section relationship. I wish I'd come across that discussion earlier!

The method that I have submitted does generate a perfect regular shape (within the rounding off limits of SU of course...does anyone know what that is?)

The key is drawing two construction "circles" and rotating two adjacent hex faces until their vertices snap on to the intersection point of the circles. Once this point is located, the rest of the polyhedron can be constructed.

It only works if the "circles" are assigned a lot of points. I used 1000 which means each circle is really a polygon with 999 edges. Then when they intersect, a point is generated which is close enough to the true intersection point to allow a perfect solid to be built.

I guess this method could be used in other mathematical constructions requiring intersections of circles.