Saturday, September 19, 2020



This is actually a couple of weeks old.  Not sure why it didn't get posted before.

I am getting some good responses by sharing progress images on LinkedIn from my work on St Anne’s Limehouse.  Interesting discussions.

First of all a pretty picture.  Current work on the ceiling: tightening up the profiles of the main cornice and the flat-panel edge-mouldings.  Then adding some detail: modillians and roses.  All highly simplified and judged from low-res photos.  I’m learning a huge amount in the process of creating this intelligent digital model.  That’s the embodied learning part.  I am deeply engaged for several hours at a stretch, figuring this building out, and it’s hell of a rush.



One point that comes up repeatedly.  Am I using a point cloud? 

Well I would love to have point cloud data for every project I work on, but most of what I do is based on much more sketchy information.  Way back in the early 2000s I bought a book in London about Nicolas Hawksmoor.  It’s one of those Thames & Hudson art books covering his whole career and well-illustrated.

A decade later I decided to visit his six London churches and take photos along the way.  Then I did that again last year.  The difference was that I had a contact at the Limehouse church who had agreed to show me around. 

But NO.  I don’t have a point cloud.  The plan I have is hand drawn, digitized from that book using a cheap digital camera about 17 years ago, and has no scale bar.  I scaled this up using an estimated length from Google Earth.  So it’s a very rough approximation, but that’s OK.  It’s a digital model so you can adjust the model based on new information and the drawing automatically updates … pretty much.



The photos are very helpful, especially the external ones.  The internal pics are a bit grainy and blurred.  Some areas are not covered at all. So what it means is that I have to do things the old-school way. Hand-Eye-Brain: that amazing interactive feedback loop that we inherited from the great ape line and evolved to new levels with our opposable thumbs and super-social brains. 

Building on traditions from Leonardo to Piranesi to Francis Ching you can learn to judge proportions, to stand back and spot relationships that don’t look right, to keep improving and refining your model.  I love the problem-solving nature of this way of working.  New insights pop up all the time. It may be about ways of using Revit, it may be about the Classical tradition, could be about church design or what churches mean in the development of western culture. 

Learning by doing.



I like to start with a very “broad brush”.  Just get something down that we can reflect on and begin the process of asking questions and forming opinions about what this building means to me.  So the truth is that I took the work I did in 2014 on trust and scaled the plan based on a family I created then. 

Time to cross-check this against current google earth data.

Well it’s not bad.  I estimate it could be up to 5% smaller than real life.  It’s always difficult to tell because, apart from the fact that the scale bar is only approximate, the photograph is never quite from directly above the building.  Here you can see a bit of the west and south sides.  The effect is particularly noticeable on the tower of course.



When I talk about modular Revit families, I usually mean a system of nested components with consistent sets of parameters that can be “mixed and matched” to create a large variety of combinations.  For several years now, I have been using this approach for doors. 

The great thing about this, is that the system evolves over time.  This weekend I decided to expand its ability to handle doorways with arched heads.  Half-round heads are easy because of the way semi-circles “snap into place” in Revit.  Shallow curves are a bit harder.  You will need a couple of extra parameters and a formula or two.

This is the way that I do it.



Now I need to apply this to my nested modular system for doors.  In the host family, we have a wall and an Opening Cut.  In this case I added some loose geometry for the plaster surrounds.  It’s not totally loose, the extrusions are hosted on reference planes, and the sweeps are hosted on the extrusions, but it’s not a nested family.  I might do that later for additional stability.

First level of nesting gives two families called “Frame” & “Doorset”  These are linked back to width & height parameters in the host family.  Nested inside the Doorset are “Panel” & “Swing” families, also with linked parameter sets.  Because these are double doors there are two of each.  The Swing family is purely symbolic, so it’s only visible in plan views, and will not be affected by the shape of the door head.



The hard work here was setting up the formulas in the nested components (Frame, Doorset & Panel) and linking together the “Rise” parameters so they can be controlled from the top level.

There is an extra complication because the frame is bigger than the Doorset, based on the thickness of the jambs.  Fortunately all the curves are concentric, so it’s not too difficult to make the necessary adjustments and get everything to flex in harmony.

The good news is that, now I have set it all up, when I want a different door panel I can just dig down and edit that component.  Similarly I can vary the frame.  It will be fairly simple to adapt this into rectangular doors with an arched fanlight, for example.  Also quite easy to convert it into a single door.

For now, though I just have the door I need for this project, and if I need to adjust the width or the height … no problem.

Wednesday, September 9, 2020


 Can we use BIM to sketch ideas out, like we do with pencil and paper?  Quick sketches, exploratory diagrams, analytical studies?  That’s part of the motivation between my invented term “the BIM pencil”.

My second slogan. “the way we build” tries to capture the idea of using a wide range of tools and techniques to understand a crucial element of human culture.  We have been building shelters for tens of thousands of years.  Cities go back perhaps 10 thousand.  There is a rich story that can be told around those different ways of building.  I am a great believer in “learning by doing” and a proud tradition of architects who studied the glories of the past using the tools and methods of their profession. 

I read and I collect books. I take photographs and store them in digital archives that I have maintained for a couple of decades now.  But I also use this source material to sketch and model.  I puzzle things out for myself.   

·         “How does this thing work?” 

·         “Why was it built like that?”  

·         “What was the construction sequence?”

St Anne’s Limehouse is my current obsession.  I started about 3 weekends back and I’ve been sharing my progress on LinkedIn as mentioned in a previous post.  In this piece I want to describe my process, which is a question a couple of people have asked.  It’s not that different from the way I approach any Revit model.  Perhaps you will find it useful.

I start with Primary Elements: walls/floors/roofs.  Just roughing out the main masses.  Actually, I always add some grids at this stage.  They will probably evolve over time but it’s good to have something that extends through all levels that you can measure back to.  This early stage is very rewarding as you start to develop a memory-map of the building, as we all do with buildings we live in, or use regularly. 

Secondary elements require loadable families.  Sometimes, (reluctantly), I opt for in-place modelling.  I learnt long ago that the system tool “Wall Sweep” doesn’t work well for classical cornices.  The problem comes at the junctions and wraparounds.  These always tend to lose their mitres as the walls that host them inevitably change.  So in-place sweeps are more stable in that regard.

I have a library of starter families for windows and doors that I have built up over the years.  Some of them are all-purpose placeholder families that can stand in for any rectangular or round-headed window at an early stage.  To be replaced by more detailed representations later on.  St Anne’s is an interesting composition.  The articulation of the main volume is achieved by recessing the walls slightly in both plan and section.  This simple, stark approach to form is typical of Hawksmoor. 

The quoins at major corners provide a linking element between the flat bands at the base and top of the walls, while also offering some finer scale detail.  I have a library of modular columns which allow me to produce something roughly the right size and style for most classical buildings.  We can do fine tuning later. 

It’s a bit like painting.  Start by blocking out the main forms with a broad brush, and gradually build up the composition, layer by layer.  Don’t get hung up on the details.  They will come to the fore later on.  Keep everything in balance, at more or less the same level of development.


The tower at the West End is obviously a focus of attention and the most complex geometric form.  So it’s worth putting in a little effort to develop this next.  I haven’t spent much time on the top portion with its octagonal array of clustered columns. But we needed a clock, an elaborate cornice and a wall-hosted family for the belfry louvres with nested classical columns (square and round)

Are these Corinthian?  Mine are … but on closer inspection, Hawksmoor omits the scrolls and cauliculi.  Just the two rows of acanthus leaves topped by an abacus.  I think there’s a hint of egg & dart there, at least in the square versions.  But that’s detail, which I promised to ignore for now, so I just used my bog-standard Corinthian.  It can be updated later.

When Rufus showed me around a year ago, we ascended a spiral staircase within the thickness of the wall.  I was completely disoriented of course, but in hindsight I realise this stair is on the South side of the tower, marked by a vertical row of slit windows. 

Then we get into a phase of tackling some more composite families, with nested pilasters, recesses,, mouldings.  I’m just doing enough to represent Hawksmoor’s scheme at this stage.  They will all need a second pass, and probably a third … if I ever get there.

How about some site context?

I did an urban study of Hawksmoor’s six churches some years ago, so I just linked this into my church file and moved it to the right position.  These buildings and their very different settings make such a wonderful group for comparison purposes.  I’m so glad that I got around to hunting them all out, and researching the likely nature of their surroundings 300 years ago.  I wonder if there is a Hawksmoor society? 

The internal organization of the building needs to catch up a bit next.  There are stairs leading up to the upper galleries on both sides of the main entrance.  There is a linking passage on the ground floor, at right angles to the main axis, and I think another just above this.  I am looking to Rufus to give me some feedback on that, and maybe some dimensions also.


With that in mind, I prepared a study sheet with questions and assumptions for him to comment on.  The door to the spiral definitely leads off to the side, half way up a flight of steps, I remember that, and I have a blurry photo to confirm.  Just how the spiral relates to the cross passage at the upper level is pure conjecture.  There is definitely a window overlooking the circular entrance lobby, so I guess that is located on the upper passage.  But I’m sure there are subtleties that I’m missing. 

The organ is backed by a giant niche, (or apse) that extends to ground floor level, forming another lobby space: half-round this time.  There is a timber screen separating this area from the nave, with half-glazed doors.  For the moment I have represented this using a curtain wall. 

I love the way this timber work echoes the organ above, and spreads out horizontally, via the galleries to embrace the whole church.  Balance of materials.  Very nice. 

That’s enough for one post. To be continued