Sunday, May 20, 2018


I think it's fair to say that churches are very prominent in the history of Western Architecture.  Clearly they are deeply loaded with symbolic meaning.  I was brought up a Christian, but the "plain meeting room" kind, and I've been an atheist for at least 45 years now, so what meaning would I find in a church building?  Why would I set about modeling and analysing a church? 

I can look at the technology, materials & structural forces.  We could think about response to climate (I guess, although that aspect is probably less relevant that in a dwelling) and of course we can look at aesthetics in a fairly abstract way (rhythm, proportions, symmetry) but without meaning it all falls a bit flat.  Music is much more interesting when it evokes an emotional response.

I carried out a series of studies of Hawksmoor's London churches some years ago, but that was handled at the Urban Design level and strongly influenced by my excitement about rediscovering the city that I first fell in love with as a student ... 45 years ago (again)  More recently I modelled on of John Soane's three London churches.  Here the motivation was to take on a Soane building that I might actually be able to finish (lol) and to compare different aspects of his style and design approach.

I've been modelling the church of San Giusto Nuovo in Volterra.  This time it was about doing something tangible with the point cloud and photogrammetry information that we acquired during the workshop I recently attended.  But of course, the more churches you model, the more you start to aske questions and make questions and as a human being it's pretty much impossible to avoid looking for "meaning".

This is the first time I have used a point cloud to inform my modeling.  It's been a wonderful experience, not perfect, but so much better than fumbling around with the kind of source material I usually start from. 

One little trick that I developed to combat the sluggish response when trying to model on top of a point cloud.  Let's say you are in an elevation or section view and you want to model a particular decorative element.  Take a reference dimension based on points that you have already established, then use the snipping tool to save a jpeg of the area in question.  Switch over to family editor using a suitable template.  Paste in the jpeg (I wish you could just paste into Revit from the clipboard) and scale it based on the reference dimension.  Now you can trace over the reference image without Revit trying to snap to individual points within the scan data, or struggle to regenerate the screen as you zoom in and out.

San Giusto appears to be the Italian version of Saint Justus, a fairly common name.  "James the just" comes to mind, "just" to prove that I have a residual biblical memory, and via this sentence, that words "speciate" into multiple branches of meaning, simply by being used.  "Ius" was a Latin word for legal right, it came to mean correct, or precise and you can perhaps see how that slipped sideways into "only".  "Just So" stories started off meaning exactly so, but by being applied to rather fanciful tales flipped its meaning so that "Justice" became "Fake News". Which I suppose is poetic justice. Words are slippery.

There are smaller scrolls to terminate buttress walls down the sides of the building.  These line up with the internal arches that divide the nave into three bays, before the crossing.  Once again the point cloud was a wonderful tool for fine tuning the size and shape, but less so for articulating the form itself.  The yellow stone used here is quite soft, so there's a good deal of erosion and patching up.  Captured Reality can be quite irregular.  I choose to model an idealised version of the church.  Purified or justified perhaps. 

So the aim is to capture the design intent, to evoke a scroll in you mind's eye.  The size and proportions will be very close, but in Revit the ramped end of the spiral is always going to be a bit of a nightmare.  So I will cheat.  The starting point is an extrusion, drawn in a side view and adjusted to match the scan. Then I add sweeps, picking the edges of the extrusion for my path and using a profile a bit like a turned baluster. 

I wasn't able to do this "all in one" so I split it into three sections which proved useful later on.  I've done something similar before with the Ionic capital.  You can add two cylinders then extend the edge curves so they disappear almost tangentially into the sides of those cylinders.  The result is quite convincing and your brain does the rest.

One of my goals for the workshop was to use photogrammetry to capture sculptural detail and embed this into a Revit family.  Recap photo was struggling a bit on site, probably down to the flaky internet, but Dave Dreffs took some of my efforts and converted them successfully.  Here are some of the source photos.  It's a coat of arms taken from the plinth of one of the side altars that line the nave of the church.  I am assuming this has to do with local patrician families displaying their piety/investing in the afterlife.

Recap photo stitches these up automatically and spits out a textured mesh.  It can be quite detailed and quite heavy.  In this case an obj file of around 100mb.  Dave also made a nice little video clip, which I think you can do in Recap Photo itself.  I've been a bit lazy about spending time in these apps to be honest.  Or to put it another way ... impatient to use the results within Revit and get on with analysing the building.

3d Studio Max will open the obj file.  I am a total novice here, but you can then go to Modifiers/Mesh Editing/Pro Optimiser.  This will allow you to decimate the mesh, basically reduce the face count.  You can also do this in Recap Photo, but you are going to use Max to convert to DXF anyway.  You need to be pretty drastic.  The DXF version we are using (2004) has a limit of about 32k vertices.


Next step, is a process I have described before where you can select all the edges and make them invisible.  If you don't do this the mesh is going to look really awful in Revit.  To put it another way, by hiding the edges in this mesh you can have edges switched on for normal Revit geometry without this shield turning into a black blob because of the thousands of visible edges.  Another workaround is to set the line colour for the mesh to a pale grey, but this is a poor second best.

Here is my mesh brought into the altar family.  The black edges really help to define the mouldings around the plinth, but the lack of edges within the coat of arms disguises the fact that is is actually a mesh of tiny little triangles.  Revit does actually create the occasional black edge where the geometry is strongly undercut, but this is a bit unpredictable and changes as you rotate the view angle.

Ultimately I made the coat of arms as a nested family and created another one that can be swapped out with a type parameter.  This is just a simple lozenge shape which is based on the reality at San Giusto.  Perhaps some of the entities sponsoring the altars didn't have a coat of arms, or maybe they were taking the economy route.

To get the coat of arms to show up nicely in a hidden line section view, you will need to add some symbolic lines.  The whole process is quite time consuming.  Minimum of twenty photos from different angles, process them all in Recap Photo, decimate the mesh, hide the edges and convert to dxf, embed in a family (scale to suit) set the material in object styles, add symbolic lines/masking regions to taste.

But if you want to model historic buildings like this in Revit it does add a layer of reality that is difficult to achieve any other way.  We've already talked about fooling the eye/brain.  To be honest, all drawing/modelling/artistic representation comes down to this.  Whether you are Michelangelo, or Marvel Comics ... you are using sleight of hand to trigger the brain's pattern recognition system into believing a "just-so" story.  Of course, Enscape helps

In the end you want to do "just enough" to convey the message you intend.  In this case it is a question of finding a balance between abstraction and reality that will allow us to ask useful questions about the building: how it is built, what it meant to the community that used it, how does it compare to say St Peter, Walworth built by John Soane a century or so later.  My initial interpretation of the bays was that the arches extended into cross-walls carrying beams that run lengthways along the nave.  But I think the span is too large and the continuity of the roof void impeded.  Who knows?  It would be wonderful to venture up into that space.

So I have reverted to a trusses above "shell vaults" interpretation.  We did see lots of examples of ceiling vaults built with the shallow flat bricks that predominate locally. Again, I would love to see Tuscan tradesmen constructing these vaults.  One of the advantages of building a model from scratch with software like Revit is that you can create it "as it used to be".  Whereas a point cloud just captures a moment in time with whatever alterations and clutter have accumulated over time. So I restored the high level windows along the aisle, on the assumption that they used to exist.  Another open question, and there are many more to be asked.

So that is my version of San Giusto, Volterra as it stands today.  I would love to explore some deeper questions.  I haven't mentioned the question of the bell tower which seems to rest partly on the edge of a vault, nor the fact that I have omitted the lean-to structures either side of the "east end".  These look to be later additions, but they could be altered versions of rooms that were there originally.  The church was designed in 1628 though for some reason not consecrated until 1775.  The style puzzles me a little.  It is often described as Neoclassical or Renaissance, which I take to be generic terms.  The date would put it in the Baroque period and it does strike me as a kind of very simple, "rustic" version of Baroque.  But those are discussions for another day. Much to be done.

Sunday, April 8, 2018


Too long between posts, and especially behind on reporting my Bank of England
work.  We've been plugging away steadily, but even with a three day weekend, time seems to run out before I can document the progress.

Waiting Room Court was due for an upgrade, including the loggia along the northern side.  Soane had a strong liking for processions of tall narrow arches.  We have come across this in his church at Walworth as well as the central corridor of the Board of Trade.  I think the loggia is one of his best variations on this theme.  The drama is heightened by the way it proceeds from the relative darkness of the Doric Portico, opens up into the courtyard, then closes up again.

In preparing this post I came to another of those "realisation moments" that make this work so addictive to me.  The paired, shallow, flat columns of the distinctive north façade of Waiting Room Court are not just one of Soane's mannerisms.  They are their because of his wish to combine an "enfilade" of arches down the length of the corridor while keeping the space open to the elements on the courtyard side.  On the internal side, the cross-walls that host the arches are framed by a pair of half-pilasters.  On the open side these half-pilasters wrap around to be expressed as those distinctive flat columns that make such an impression in Yerbury's record photos from c1930.

Another subtlety I quite enjoyed was the splitting of walls in section view to create the hint of an entablature resting on the pilasters.  Along with the two types of arches, come two arrangements of ceiling coffers, both quite elaborate and treated here as loadable Generic Model families.  The enscape3d image below was captured before creating the coffers families and the three images on the left record a developmental sequence, starting with an in-place extrusions, copy-pasted into a family template, then cut by voids.  Is this a Soane original design, or is it developed from a classical precedent?  Obviously stepped coffers, and oval coffers were not invented by Soane, but what about this combination.  I really don't know.  Perhaps someone will stumble across an answer.

It's important to remember that Waiting Room Court (WRC) is built at the junction of old and new.  Princes Street used to run diagonally across this space, until Soane straightened it on behalf of the Bank as it flexed its muscles over its former landlords the Grocers' Company, who owned the land where the loggia was built.  So the loggia represents the new regime: modern banking with shareholders and a board of directors, while the other 3 sides of the courtyard recall the older world of medieval guilds and hereditary occupations.  Whether consciously or not, Soane has expressed this transition in his architectural treatment.  And it's interesting to note that he maintains a balance between contrast and harmony.  The courtyard holds together despite the rather different elevational treatment of the loggia.  All four sides stand on the same rusticated base.  Britain managed its transition into modernity without the cataclysm of the French Revolution partly due to the moderating influence of the Bank of England.

It's almost embarrassing how easy it is to capture compelling images with Enscape3d.  Here you can see the two types of coffer and the two different arches.  Also on view is the higher ceiling in the central bay, which is still very crudely modelled.  I'm tempted to add some top lighting here, although the survey drawings suggest blind recesses rather than windows.  This is not definitive of course.  The survey also shows the narrow end bays as open, which is a change that was made later on by another architect.  Certainly it would be unusual for Soane to create this additional height without lighting it.  This was done before gas lighting remember. 

The WRC itself was long overdue for a facelift.  Window families were just simple openings in some cases.  The main cornice had been created as a wall sweep and kept coming apart at the corners.  Replaced with an in-place sweep. The basement story needed to be treated as a rusticated base so we made a start on that by fleshing out the large half-round windows.

It's clear that several of the windows are "blind" because they back on to the junction between rooms or, in one case overlap between the top corner of a vault and the ceiling void above.  To achieve the blind window effect, replace the cut opening with a void extrusion.

As the rustication proceeded it seemed appropriate to update the render appearance for the walls to a more stone-like material.  This led to some quite dramatic images, but ultimately the stone finish is not correct for the internal spaces, so I have begun to set up new wall types

All the same it does feel good to have additional richness in terms of material finishes.  We will have to look for other ways to achieve this moving forwards, maybe it will come in features like the fireplaces, and perhaps we can start to add furniture, carpets, etc. to bring spaces to life.

An interesting question that arises in my brain from time to time.  Why spend time creating seductive, dramatic images?  How does that relate to "historically accurate modelling" on the scale of aims and objectives?  My answer usually is: this is all about understanding, finding meaning in our built past.  Understanding doesn't always arise through conscious reasoning.  In fact the evidence is that our opinions are formed at a subconscious level and subjected to conscious reflection after the fact.

So I would say that it's a cyclical process, like anything in the creative realm you pour energy into an activity then stand back and review.  Emotions and reason intermingle to drive the whole thing forward.  Insights often come apparently from nowhere: those Eureka moments.  So while you are "waiting" for the next blinding insight, why not feed the brain with stimulating visual fodder?

Sunday, April 1, 2018


My trip to Volterra is looming and I wanted to do something to get me in the mood, so I started developing an approach to parametric modular windows of a particular Tuscan style.  Here is the first fruits of that venture.

The starting point was a parametric pointed arch. From there I went on to develop 3 more families with nested families for the stone "Insert" and the glazed timber "Window Unit".  The idea is to set up a modular system where you can swap out the Insert and the Window Unit for different versions without starting again from scratch.  By keeping the names constant, all the parameter linking remains intact.

I am using the nested planting family hack to easily scale the columns.  The intention being to allow for more complex capitals and base mouldings, while retaining ease of scaling.
All was fine until I checked out the windows in the plan view.  First of all the top part of the insert tends to show up, blocking everything else.  This is easily solved but I was still left with columns that don't react to the cut plane. 

The complexities of how window families interact with the cut plane are quite interesting, and I recounted my explorations of this, long ago.

As it happens, things get even more complicated when there are nested components of a different category.  Usually I keep all my nested families in the same category as the host family, so that subcategory visibility is predictable and consistent.  But the planting hack throws a spanner in those works.

I don't have time for a detailed analysis just now, but suffice to say I did a whole series of permutations to get the columns to cut nicely without success.  Strategies that seem fine in family editor don't pan out in the project environment.  Conversely families that look fine when placed directly in the project fail to cut when nested into a host window.

In the end I have opted for a shared family for the columns.  This is in the column category, with the "pre-cut" option unchecked.  Probaby a shared Generic Model would also work.  So there is an "inner" planting family, inside and "outer" planting family, inside a shared column family.  The height of the column family is an instance parameter so that it can be controlled from within the host.  You can't link type parameters for shared families because that would defeat the whole idea ... that the family acts as if it was placed directly in the project.

Maybe Paul Aubin has a better solution to this, or even a better approach to the whole business.  I'm sure we will talk that through when we meet up in Volterra in a couple of weeks time.  We won't have much time for making families during the workshop.  The focus will be on collecting data with various bits of reality capture kit. 

At the moment the families are based on a few rather grainy internet images.  It's a very interesting approach to windows I think.  Had never really thought it through before.  It's clear that there is a rectangular window with side-hung casements (must open inwards) and this is tucked behind a decorative stone "double arch within a larger arch"  I'm very keen to scan a few examples now to get the sizes and relationships right. Probably there are more surprises in store.  Internal shutters ?  Splayed internal reveals ?  Sill treatment ?

Then we could develop a range of parametric components which could be used to assemble urban settings like the main square in Volterra.  That's my basic premise anyway.  By the way, here is my solution to the pointed arch parametrics.  Didn't check out any other ideas, just decided to develop my own approach