On the Eave of Construction (II)

With weather remaining endlessly cold, my shop remains under that 40˚F threshold, a mark which I have decided is my cut off for working in there. I’ve been continuing on therefore, with my work at home to draw a certain type of Japanese residential roof, with an aim to cementing my knowledge of how to learn RhinoCad and work out some long-standing carpentry drawing issues which I couldn’t resolve with the previous drawing software.

I’ve put a lot of hours in, and things have been going well generally with the drawing work, at least as far as the CAD aspect is concerned. Essentially, I find that one has to flail away for a certain amount of time, have certain things not work out for some reason or another, and then try a ‘plan b’ and do it all again, and probably again, until eventually you can produce what you want. By one’s own mistakes in modeling, one learns better ways to model.

The reason for this is that with many given tasks in Rhino, there is more than one way to do it. Certain habit patterns in how I go about modeling parts I bring over from sketchup prove to not work so well sometimes in Rhino. In SketchUp you produce solids a lot of the time and then push, pull and intersect them to obtain the finished part you want. In Rhino, one can do all those same sort of techniques, but they may lead to a part which is a collection of faces that will not weld together to make a solid or some such outcome.

As an example to show the variety of potential approaches to a given objective there is the task of ‘creating a surface’, one of the most basic in modeling:

  • One could opt to draw 4 lines to produce the outline of the surface, then join them together with a separate command, then patch that outline with a surface.
  • Or, one could use the rectangle tool to produce the outline, then patch.
  • Or, one could draw the rectangle with a ‘polyline’ tool and then patch.
  • Or, one could use the ‘surface’ tool and in one move it’s done.
  • One could alternatively have previously created a cube, from which one can extract the desired surface, or copy that surface, or explode the cube and subtract everything you don’t need to obtain the surface you want.
  • One can take a set of lines comprising the surface one wants to model and then use a tool to loft a surface over that outline.
  • One can take the lines comprising the surface and then use a tool called ‘curve network’ to create the surface.
  • One could take a larger surface and then use a ‘wire cut’ tool to slice out the portion of surface in the shape desired.

I could go on as the above list is hardly exhaustive, but I’m sure the above list is enough. it’s the same for a lot of modeling tasks, with various ways to approach a thing presenting themselves. And when you are new at it there is a tendency to stick with a the one tool you got a mental grasp on, and use that exclusively, until you find one time it isn’t working for some reason. And thus you must learn other ways to do the task, and eventually one comes to understand why one chooses one tool over another to get a certain job done efficiently and well. It’s no different than woodworking really, save for the fact that the drawing work is virtual.

As a concrete example, here I’ll show a curved hip rafter. In cases like this one, where the curve is not of continuous radius, the ‘backed’ top of a curved hip has a shape which changes shape as it moves along. Thus the surfaces of the backed cut on each side are twisted planes:

earlier version

This kind of twisted surface is a modeling task which was all but hopeless in SketchUp. In Rhino, there are several ways to do it. In the above I used a tool which produced quite a busy surface, but it was mathematically precise.

Later, working on a similar task on the ‘field’ hip rafter, I was able to obtain the same result with a much less busy surface, which is certainly easier to look at:

looking across

I’m very much at the early learning stage with some of these tools, so a lot of stumbles still lay ahead I’m sure.

The problem of learning which tool to use and why, is one aspect that has been another. A second point of struggle relates to the layout manuals which I use. My favorite ones are by Togashi, and they are certainly comprehensive. I thought they were essentially flawless, however I have recently discovered that he makes mistakes too.

All of my drawings for roofs and such are performed much as if I were doing the task in my shop- sketching first in 2D to produce the required shapes and then using those shapes as templates to make the parts. Here’s one of the developed views I sketched of this roof model:


Towards the bottom right is where one projects lines from the view of the eave timber build-up over to the hip rafter:

overview b

I followed Togashi’s methods, however because I am able to create the parts afterwards as 3D solids and then superimpose those solids back into the 2D drawing portion, I am able to check whether the 2D method is producing parts which are shaped and positioned correctly. And that’s where I found problems with Togashi’s book:


It’s a problem I actually discovered previously and had forgotten about. As you can see I do scribble notes in pencil on the textbook to keep track of things like this.

His mistake in this case is to project points over rom one view, which are of planes along the centerline of a part, then then reflect them in a secondary view but on the wrong line of reflection. If you look at the above sketch of drawing the ‘kera-kubi’ hip nose treatment, you can see that where he has marked point ‘e’ is where he reflects a line showing the front face miter from the eave’s kaya-oi timber – but the line he is reflecting the projection upon is the line for the face of the hip rafter, not its centerline.

The consequence of the above mistake would be to cut the angled receiving notch on the end of the hip rafter, right at the point of the miter, too far uphill. The kaya-oi would bind in that location, as the error in line position is about 4.5mm. It’s not as if he would be wanting to make that sort of mistake, would he? Not generally an objective in Japanese pursuits, as far as I have noticed.

I’m not expecting readers here to grok exactly what I am talking about – just to state that there is a problem in the text that I was initially in some disbelief about and had to overcome. And it is not just that page, as he repeats the same problem on other drawings, and I have discovered other minor mistakes besides.

And I’m not trying to run the guy down either – I esteem his work greatly and I’m a little shocked to find any mistakes in it as it is that good otherwise. I think my main point  is that having the ability to construct parts in 3D solids off of 2D, and back-check them is a truly wonderful thing. I think that if you are only drawing in 2D all the time, with the myriad of lines to stare at it is only a matter of time before even the most experienced folks get a little mix up here and there.

Anyway, I have made some modest progress after a bunch more struggle. Here’s a view of the eave from underneath with the eave build up complete and rafters in place:

version 2 from below

A view from the top side shows the curved field rafters getting put in, and gives a look of the field hip rafter:

version 1 above

The fact that the field rafters are curved down, or sagged (called ‘tarumi’), means there are some complications in the production of the field hip upper lines, as one must account for both the cure in the field common rafter, and the fact that the rafters travel along an eave edge which also curves up at the end.

I had made the field hip rafter deeper in section in the lower portion so it could sister against the decorative hip rafter below. However, after I had the parts together and could take a look, I could see room for improvement.

So, I started those parts again, and this was the result:

version 2 above

Besides the improved junction between the two hips, the uphill portions have been lengthened and straightened. I’m happy with the outcome.

Another view:

version 2 below

There’s a heap of work to do yet on that sketch to complete the building. The roof is not hipped, I might note, but has hipped gables, and that curved gable is one of the more complicated carpentry drawing aspects to Japanese roofs, so it will occupy few hours I’m sure. I think once I have the roof modeled I will hold off drawing in the rest of the building, and maybe move to drawing a Japanese roof on a hexagonal pavilion or something like that, but we’ll see.

I hope at least some readers found the above investigation at least interesting, whether or not Japanese roof carpentry holds the same point of fascination for you as it does for me. Let me know in the comments section below if this is your cup of tea or not. Thanks for coming by the Carpentry Way. Part III follows.

5 Replies to “On the Eave of Construction (II)”

  1. Nice work Chris.
    In Solidworks you can construct the individual parts and place then in an assembly , then mate parts together in a variety of ways, and clearly see any interference or error. It is very easy to find any errors if fitting parts together.

    You can easily construct planes and draw from any point of view. Slice and section views. I think that you would find it the right choice for you,try to get someone to show you how it works. I have done quite a lot of work with it and find it perfect for me, i do some joinery models and furniture but also lots of mechanical stuff, some of which has quite complex shapes requiring many planes.
    I really believe Solidworks would be the one for you.If you can find someone that uses it to show you, i think you might agree.

    All the best.

    1. Hi Mark,

      thanks for your comment. While I’m equally sure that Solidworks would be a great choice, I’ve already selected Rhino and am finding it more than satisfactory, so not too likely at this stage that I will be jumping into anything else. Thanks all the same for the recommendation.

  2. Neat stuff! You are right that Rhino can be used either as a surface or solid modeler – I usually try to convince my architecture students to maintain the discipline to stay in solids if at all possible. I find surface modeling is great for conceptual sketching, but on the path to madness if you are working towards fabrication of any sort.

    If you want a taste of Solidworks but aren’t ready to take the plunge, check out OnShape – it’s an online procedural solid modeler of a similar genre. It’s free if your model is publicly available; I’m usually using it to make small models for the CNC where the procedural nature really helps with iterating and planning out complicated, multi-stage cuts. This is a wood deadeye which involved CNC cuts from four sides: https://cad.onshape.com/documents/e693b582af4f4cbdb2b8c5ac/v/0c46a66ec4bcf77d122f314d/e/4e8ed9a850f44d36af217c3b

  3. I did some simple mechanical drafting in college years before the PC was invented. I have always been amazed at how it can focus the mind and reveal secrets that are oh so easy to mentally skip. I love seeing, and reading about, your efforts. I am just barely able to follow the gist. However, I love both Japanese and Chinese roofs for the sheer beauty and even small bits of understanding of the complexity behind them only enhances that. An overly wordy way of saying keep it coming.

Anything to add?

error: Content is protected !!
%d bloggers like this: