A Sketchy Situation

Draw first, then build – seems simple enough in theory. Drawing is best informed, I would suggest, by how the item is to be made, and this is where a person who makes as well as builds has a inherent advantage over those who only design. That said, one caveat is that sometimes new and original designs come about from those who do not labor under preconceptions and strictures imposed by best practices in a given field’s fabrication practices. It is all too easy though to design things which are impractical or even impossible to build, and these sorts of designs generally emanate from those who do not get their hands busy with actually making.

Division of labor of course has it’s benefits, but it is not without drawbacks.

The means by which one draws is itself a controlling factor in regards to design. I thought it might be relevant to some readers if I recapped my own approach to drawing over the years, providing some explanation for how/why the process has evolved in the direction in which it has, and why I have made some recent changes.

A long time ago I drew on sheets of paper or door skin, either to a scale or 1:1, using the conventional tools of ruler, compass, pencil, etc.. Those methods remain valuable to this day, as sometimes the most straightforward approach, especially in regards to larger architectural items, is to produce full-size templates on door skin and use those to transfer lines directly to the wood.

With furniture-scale pieces though, I’ve found in recent years that I save a chunk of time by having relevant portions of my CAD work exported to a large plotter which makes full size paper templates, which are then cut out, applied to door skin or thicker substrates, and used for layout or as a cutting template.

And my switch to using the computer to draw was prompted by a study I did a long while back in how to manually draw an item in perspective, using station points, and so forth. I found that the sheer number of traces required was leading to problems on the paper with lines which were really close together and thus in pencil or pen resulting in a big fat mess by the time one was halfway along.

These problems have been solved in the past by folks of course, yet one cannot deny the astounding amount of time it takes to prepare complete perspective views, especially of more complex objects. It’s time measured in hours and even days.

Drawing in perspective though is more or less essential when it comes to communicating with clients about what you propose to make for them. I’ve had experiences where I have presented the usual plan and elevation drawings to a client and had them nod and look like they were comprehending what they were seeing, but later found, through client comments when the work was well along, “oh, now I see what you had shown in the drawing”. A lot of people, it turns out, do not relate especially well to standard plan and elevation drawings, let alone more involved stuff.

And even from the point of view of being the designer, the standard views connote a limited amount of information. It’s hard to really get a sense of what the piece will look like when you are standing there next to it, or looking at it from across the room, say, without some sort of perspective rendering.

Also, drawing on paper or other physical surfaces tends to engender a ‘one-shot’ approach, especially for those sketches which take a while to complete. While lines can be erased and redrawn, if one has to do too much of this then things get messy. The more invested one gets into a sketch on paper, the more one tends to cling to progress realized and to loathe making significant changes. And of course this severely limits exploration on the design front.

Funny enough, a similar kind of thing happens with drawing software too, as hours spent invested in learning one type tend to disincline one from starting the process over with another software.

After encountering problems with drawing in perspective on paper, I decided to give Computer Aided Drawing (CAD) a try. The first one I opted for was called MacDraft by Microspot, a 2D drafting program. It is still on the market in fact. One of the first things I tackled was to employ the traditional means of station points to produce a perspective view of a heptagonal splayed stool. Here’s the finished product, with all the drawing traces removed from the view:

I don’t think I could have accurately drawn that piece like that were it not for the formal method, and CAD in 2D made it happen for me. It was immediately apparent that 2D CAD helped me do things that were difficult to realize with pencil and paper. It’s easy to modify digital drawings in most cases, Still, it took a long time to put all of the traces in there so as to be able to produce a perspective sketch.

A few years later after taking an introductory class in French Carpentry drawing from Boris Noel, I was engaged in a detailed study-draw-build for the infamous 19th century Mazerolle tréteau. That piece was rather complex, and my study was hampered by having neither a teacher nor a physical example of the completed piece which I could examine and learn from, and oh yeah, no fluency in 19th century French. Most of the traces required to draw all the parts are absent from the sketch, by necessity, in the Mazerolle book Traité Théorique Et Pratique De Charpente. The backside of the drawing of the piece in perspective was not visible, so I was left unsure of how the connections between parts looked when brought together on that side. I went as far as I could go in 2D before the doubts and uncertainties about what I was drawing, in other words how well I was understanding the method and whether it would actually produce the required parts correctly. It seemed like folly to try and construct the piece until I was sure that I had a handle on whether  things were working out correctly in the drawing.

This is about as far as I got with 2D in MacDraft:

This drawing work took place around 2008. The nagging questions about whether the drawing method in the book which I was trying to emulate was producing the right parts or not led me to my first 3D drawing program, SketchUp (by Google). The program was free, and users could submit their drawn models to a warehouse where others could make us of the files. Things in the warehouse like appliances, fasteners, etc, which otherwise would soak up crazy amounts of drawing time, were most valuable to me at times.

SketchUp was fairly intuitive to learn, pushing and pulling shapes around for the most part. I’d looked at a few other programs, including Autocad Lite, but as far as dipping my toes into the pool, Sketchup was easy to go with, and free does help make the move risk-free.

And the program allowed me to construct the traditional sort of 2D sketch on the virtual floor, and then erect the parts directly above, using plumb lines off of the developed views – hopefully this sketch gives the idea:

I remember a rush of excitement the first time I drew in the scissor braces in 3D and found that the 2D geometry was producing the correct parts so that I knew I was on the right path. Later I was able to identify more than a dozen errors of various sorts in the original illustration, and I’m not sure I would have caught those otherwise. The shift to 3D allowed me to leverage uncertain 2D developed drawing work into something that I was sure worked, and this lead later to actually constructing the piece.

SketchUp came out later with a paid version called Sketchup Pro. While I continued on with the free version for several years, I eventually opted to obtain SketchUp Pro, for $500, for a job in which construction blueprints were required. SketchUp Pro includes ‘Layout’ which enables the production of various size formal construction documents, like this page out of a blueprint set I produced for another contractor for some standard kitchen cabinetry.

In order to communicate with the architects upon whose drawing the above work is based, I had to get SketchUp Pro. Since then I have barely used Layout, and for most users I would not have recommended they pay for the Pro version since you really don’t get much more than Layout for your money.

I found that even with my new paid-for SketchUp Pro, all was not ideal. For one thing, it is supposed to provide the ability to both upload and export drawing files in AutoCad format, the standard by which most other programs on the market are measured. File formats that AutoCad works with are .dwg and .dxf.

I found that if I import an item into SketchUp which is in an AutoCad format, it will import and look fine, except it will often not be defined into different planes and discrete parts. What I mean is, if you click on the drawing, click the containing box open, and the ‘select all’, the entire part is highlighted:

This means that, unlike separate parts in the real world, like the fixing bolt and handle casting in the above sketch, are not separate but form a continuous surface with each other. It’s not exactly easy or quick to extract out the parts from one another in the drawing either. So if you have to modify one of these type of sketches in SketchUp, well, you learn it is best to avoid such tasks if at all possible.

That’s an issue with importing in SketchUp from AutoCad files. I’ve also had to export .dwg and/or .dxf files from SketchUp to a cnc fabrication place in the past, and they found that while the files I sent out of Sketchup (in the .dwg/.dxf file formats) could be uploaded, they also required hours of work to make usable for their machines.

It seems to me that SketchUp’s capabilities in terms of handling AutoCad format material is at best imperfect.

SketchUp’s less than scintillating performance continued with how it deals with files that grow in size to the point where there is a high count of polygons. I’ve encountered this a few times with different drawings. When a given sketch gets to a certain level of complexity, then SketchUp gets sluggish and even crashes. Crashes where I need to restart my computer. Later they introduced ‘Bugsplat Report’ to deal with sending info about the crash to the company – SketchUp now acquired by Trimble. It seems the company anticipates that program crashes are going to be sufficiently frequent to merit a special reporting application.

Trimble initially continued with the set up of there being a free version of SketchUp and a Pro One. However, when new versions were released, they charged for the update. Then they came up with the wondrous ‘Maintenance and Support Subscription’, and every year there would now be an annual fee to keep current. If your account was current, then you were eligible to receive any new versions of the software. if you let things lapse for a few years and then found yourself needing the current version, you would have to pay for all the updates that had occurred in the interim, if I am not mistaken.

If you didn’t keep up with the new versions, saving your cash – and why would you since the upgrades between versions, say SketchUp 2016 and 2017 were minimal to say the least – then you could continue on with your older version. However, they have made access to the 3D warehouse contingent upon having an up to date ‘Maintenance and Support Subscription’, so if you need to make use of that feature, you would have to pony up.

Ponying up for the ‘Maintenance and Support Subscription’ was around $99 at first. Then it became $99 if you paid early, and $129 otherwise. Last time I paid for the renewal it was $120 even though I had paid early, what would have been $199 otherwise, IIRC.

A week or two back I got an email from Trimble stating that there was a new 2019 SketchUp available for download. I clicked on the download and while it uploaded I took a look to see what features, if any, were new.

In that sort of obnoxious, breathlessly-excited language, we have SketchUp “Proudly introducing SketchUp subscriptions.” Dah-dah-dah-dahhhh!

Here’s how that awesome groundbreaking development looks:

In “challenging themselves to earn my business every single day”  – how nice of them to put themselves on the line like that, and all for little old me – they’ve gone to the same subscription business model as has Autocad. No more free version now. Pay as you go, and that will be $300/year going forward. Of course, one could also pay $1200 to have their ‘Studio’ version, where on offer there is “Everything you need to design better buildings’. I think I’ll pass. In fact, I’m not sure I will even bother to open up the download for SU2019.

There was one improvement of ‘note’ in the new release. Sketchup had at last added a capacity to add dotted lines to a sketch:

This is about how pathetic the “exciting new upgrades” are. Long requested – yes, but this took over 10 years to add? Are you freaking kidding me? The release notes for the 2019 version mention numerous changes to .dwg/.dxf file import and export, so it seems at least they are aware of those issues, and maybe that will work better now, but I’ve now lost interest and become disgusted with them. Sure, maybe some nerdy guys spent months making little tweaks here and there, but, weighing this with how little has changed or been improved in general with this software over the past 10 years or so, I’m nonplussed at best.

Looking through the public comment section of their splash page on SU 2019, I found I am not alone in being disappointed. Indeed, some of the comments were quite scathing in their critique, and I wanted to share some of those here, but lo and behold, upon returning to the page I find that they have now removed that comments section entirely from that blog post. Well, hardly surprising to have that section on there to dampen the hype in the message you want to get across. I find that humorous, if not a little disappointing.

Anyway, that was one of the last straws as far as my relationship with SketchUp goes. The proverbial nail in the coffin was provided with a recent and all too-familiar struggle to get a part of a drawing to work for me.

Now, there are things that take a while to draw and yet are quick to execute in the shop on the material. There are also cases where something is quick to draw but time consuming to do in the shop. Chamfers are a case where either condition can apply, and chamfers are something inherent to every woodworking project, even if little attention is paid to that aspect by some.

In SketchUp, unlike most CAD programs, there is no special tool for applying arris treatments like round overs or chamfers. You have to create a ‘cutting plane’ and then use the ‘follow me’  tool to manually swing the cutting plane along the arris which you want to treat. Once you’ve swept that plane through, in some cases you have to manually subtract, by clicking, the parts off which you don’t want.

One thing you will find, if you want to chamfer anything but the simplest shapes (i.e., cubes and other parallelopipeds), SketchUp can struggle, and nowhere more so than with chamfers following along curved lines. God help you if you want to intersect curved faces. With smaller parts, like a stick on the order of 1″ x 2″ (25x50mm) in cross-section, the the tool will simply not work without causing digital devastation. It seems that once the curve radius goes beneath a certain size, problems arise.

A commonly suggested and sometimes useful trick for this problem is to enlarge the size of the part and retry the ‘follow me’ tool. I tried this, enlarging the parts in an entirely separate drawing to 100 times the original size.

I thought I would walk through this to illustrate the issue better. Here’s the end of one stick with which I would like to explore chamfering treatments:

I created a small bevel shape and follow the procedure along the nearside arris, however after the move is done, suddenly some of the adjacent faces disappeared:

Then follows several minutes of tediously clicking lines across these unexpected openings in an effort to reestablish the faces – here you can see the top face starting to come back after 2 lines have been dragged across, though notice too that the mortise opening has also been capped in the process, something I will have to undo afterwards:

Similarly, the side face can be reestablished by snapping lines across and deleting them afterwards:

In some cases a single line across makes the entire face reappear, but in other cases it takes many lines, and sometimes no amount of lines will patch in the face. In some cases I find that after i have snapped the lines across to reestablish the face that removing those snapped lines causes the face to disappear again. Sometimes I find the face has been made into doubled or tripled surfaces, and I have to click these extra faces away.

The part above was created in a fairly typical way, and the nose treatment was shaped by drawing the profile and extruding it into a solid, so I’m not sure where the error from my side might be. It could be me, but I think it is the software or possibly the interplay between that software and my computer. Hard to say. I even added a plug in called ‘Clean Up’, which is there to get rid of superfluous faces and areas where lines aren’t connecting, as so on, and that has helped, but not 100% of the time.

In this example, after a while of mindless time-consuming clicking, I had closed up all the faces except this one on the bottom, which remained stubbornly unwilling to close up regardless of how many lines were snapped across the gap:

I just hate wasting time with stuff like that. All I did was use the ‘follow me’ tool to sweep a profile of the chamfer along the edge, and with the faces suddenly disappearing around it, I’m left with a bunch of idiotic work to close things up and get back to what I should have had after the chamfering operation.

I’ve found that sometimes the use of the ‘follow me’ tool results in slight distortions to the part you are trying to chamfer and this cause all these glitches with the other faces – but why does it happen at all? It’s ridiculous.

If 100x magnification of the parts did not lead to a good outcome, then I thought, why not try 1000x? Ridiculous as that may be, I did try that too, to no avail. Same problem.

Now, am I the only one experiences weird problems with the ‘Follow me’ tool? I think not. Now, for sure, some people’s problems with the tool attribute to basic errors in use, as they might be new to using it, but I have been using this thing for years and it works on most stuff, but not all.

The above example, by the way, is just one of many in which I have encountered goofy problems in SU, whether with the ‘follow me’ tool or via the intersection of curved planes in various situations. Good luck with that, by the way.

So, with the freezing temps limiting my shop time of late, I decided to invest some hours into my Rhino 3D CAD program. After some research I’d settled on this as a next step in drawing software, and thanks to my wife’s educational pursuits I was able to obtain it at a significant educational discount.

Now, the weird thing is that I had purchased Rhino 5 for Mac about 18 months back, and had invested some time into learning how to use it and had produced a hip roof model – here are but some of the bare bones of that:

But when I returned to have another stab at learning Rhino, I opened up that previously done drawing and found I could not remember very much about how to use Rhino and was pretty much baffled at how I was able to produce the above sketch in the first place. Discouraged, I closed the drawing and continued on using SketchUp, despite all the teeth gnashing that went along with that.

However, this round I decided I would push on through the early quicksand and dread and get a handle on using Rhino. 30 or 40 hours later I am starting to feel comfortable, and don’t believe i have any need to return to SketchUp.

Let’s take the example of the same sort of problem as shown previously (chamfers)  upon which I had found Sketchup to be extremely aggravating in use. Drawing the nose  on a stick (at nominal size, not enlarged by 100x), with a treatment slightly more involved than the previous iteration, took just a couple of minutes:

Now let’s try a 1/16″ (1.5mm) round over, or fillet -click, click, click- done:

A 1/8″ (3mm) fillet:

A 3/32″ chamfer:

It’s just a few clicks in each case to produce the different varieties of chamfer. The parts can be easily copied and arrayed side by side so one can compare the different treatments. I find this liberating frankly after my previous struggles to do what is otherwise a simple task. I’d rather be able to model it quickly in various iterations than have to resort to mock ups in the shop, though there is something to be said in favor of that approach too if you have the time.

After some 50~60 hours of diving into Rhino, I am working on a couple of Japanese timber frame roof corner drawings, and still very much a newbie with the software but it is getting me where I need to go:

I do use Rhino, just like SketchUp, in what is likely a rather atypical way at times, doing 2D developments and then creating 3D parts from those, and I’m glad to find it works for me. Most of the time though, if I am drawing furniture, none of the 2D developments are required and I draw pretty much like most other modelers. I suspect.

I do though feel I am over the initial portion of slogging in the learning curve where I at times feel completely dumb and unable to do much of anything with the program. I think the start of the process really is the hardest section, getting past the wall of the incomprehensible. It really doesn’t take all that long or require special abilities – you just have to be determined to get through.

And, like SketchUp and other modeling programs, it is not necessary to be fluent with every one of the tools available, though that is certainly advantageous. You can do a lot of modeling with a fairly limited repertoire of techniques. It’s just like woodworking in that regard I suppose.

Others with more experience with Rhino can show some of the capabilities there are with the program when it comes just to the topic of chamfers:

That’s just the basics.

Here’s a modeler refining just a fillet corner intersection by just one method of several, each giving subtly different effects:

I find the use of the tool at the end quite cool, where the cube is given a mirror surface so that one can examine how surface distortion associates to slightly different ways of bringing adjacent planes together at a fillet.

Here’s one more, fairly short, showing more on dealing with rounded arrises when parts are intersected, and again the reflective surface tool is employed at the end:

In this post I chose to talk about a relatively minor design aspect, that of chamfering, just to illustrate an issue. Rhino is capable of so much more of course. Rhino is going to be a lot better for curved forms of any kind, of that I’m sure. I think the first 8 hours in ‘the pool’ are the worst, but once you get a handle on a few tools and used to the command-based operating system, it works really well. And you can push and pull, slice and dice surfaces just like in SketchUp if you want to model that way. And unlike SketchUp, it is seriously accurate. Looking through the options for drawing units in the general settings, for example, one sees even angstroms listed. Are people drawing semi-conductors and other computer chips in Rhino?

Rhino is hardly the only choice out there and as my experience is limited in regards to other drawing programs I am not here it make any sort of recommendations, just relate my experiences. If you want a free program that can do a lot, working with polygons as does SketchUp you may wish to look at Blender. There are many others besides. Among the paid options, some are likely too expensive for the lone artisan – like Solidworks. AutoCad has some cool looking stuff like Inventor, but again not so cheap and all of them are moving or have moved to the subscription model. If you have a design office with a dozen people drawing at computers then, say, a $5000/year cost to keep current with software is not unreasonable I suppose, and this may even be true if you are a one-person business devoted exclusively to design work. Cheaper than running a woodworking shop for sure, hah!

If you, like me, draw so that you can build, then certain costs are hard to justify for the CAD aspect. Rhino, at a regular price of $695, seems a reasonable mid-point in terms of price, and seems to cover most of the bases in terms of functionality. And, as it states on Rhino’s site, “All licenses are permanent and do not expire. Prices include support and service releases. There are no maintenance fees.” There’s something to be said for that.