Thursday, December 31, 2020


 More reminiscing. This is a continuation of a thread from mid October.

In 2013, one of our lead designers at GAJ was a Spanish architect called Ignacio.  He was responsible for my spiky geometry explorations.  His team were all Sketchup guys, but we often talked about how to capture the geometries he was exploring within a Revit environment.  One of these projects was a diagrid flowing over a “rolling hillside” form.

I learnt quite a bit more about divided surfaces and complex geometry.


Furniture.  These days I am often involved with bringing the Interior Design component of a project into Revit.  Usually my role is content development and designing “BIM processes” ... schedules, data, specification codes.

I have quite an interesting collection of furniture content, acquired over the years.  One file is devoted to designs by Fritz Hansen.  The approach to generating the required geometry is quite varied.  Sometimes it’s best to download a mesh from their website and combine this with symbolic lines and masking regions in orthographic views. (Some people reject this approach) Sometimes native Revit geometry does a good job.  Sometimes you need to be a bit more creative with the Revit geometry tools.

These days I usually use very low-res geometry for the 3d views, and a CAD-like representation in Plan.  The visualisation and detailed spec is handled by a non-BIM team.  We just need a recognisable object which cross-references the specification code.  It’s less than ideal in my view, but that’s the current state of play in my office.


I was exploring “the space of possible rigs” following on from the success I felt I had with my “scalable rectangular rig.”  Presumably the simplest type of rig would be a straight line.  What can you do with that?



The rectangular rig is amenable to expressing the graph of an equation, could be a quadratic, for example.  I explored this idea to create a general family of forms, of which one particular instance would be the Gherkin by Foster Associates. This is based on work I did for one of my presentations at the Revit Technology Conference in Auckland a few weeks earlier. 

I’d like to paraphrase the last sentence of this post here.

“If you are interested in the History of how buildings work, then a tool like Revit is an amazing aid to research”

Seven years later we have developed that idea much further with our work on Project Soane and of course the Notre Dame model.  Open-source, collaborative research into our built history.


There follows a series of posts that take the scaling properties of the Planting category, and nest it into system families like Curtain Walls.  The first one is based on a common sand-cement “brise soleil” block.  How useful is it to make this infinitely scalable?  That’s a moot point, given that these things usually come in a couple of standard sizes. 

Maybe if you were designing screens made from cast aluminium panels.  Depending on the quantities you might be OK with a custom mould for the project.  So at design stage, the ability to easily vary the scale of the pattern might be useful.




The next post takes the scalable, decorative curtain panel concept a bit further.  At least it imagines further examples of designs which might benefit from this approach.

“I'm backing the double-nested technique for situations where you want to play around with the size on the fly, experiment with design ideas quickly.”


Back to classical architecture for the final post in this series.  Stone balustrades, carved mouldings, standard designs like egg & dart.  Can the scaling properties of double-nested planting help out here?  I made quite good progress with this here.  And since then I have developed these families a bit further.  In the comments Paul Aubin raises an interesting point.  How do you do this around a curve?


One answer I explored later was to use a railing.  It’s not very nice to have carved stone enrichments, (or maybe plaster cornices) in the Railings category.  If only we could assign a railing to a couple of other categories.  Maybe Wall Sweep, or even Generic Model.  I’ve no idea how realistic it would be to do this within the Revit core code, but it’s the kind of thing that I think would be useful.  One of the annoying “restrictions” we would like to remove, to paraphrase a recent post by Tim Waldock.



Tuesday, December 29, 2020


There are several areas of the Bank of England where the bottom story is treated as a rusticated base.  So far I have been “cheating these by adding a thin extrusion created by drawing out the pattern of stone blocks.  It’s all a bit tedious but that was the choice I made.  In practice these kinds of treatments often have splayed joints which emphasise the solidity.  I decided to experiment with a different approach.



Text from a LinkedIn post I created at the time ( a couple of weeks ago)

My day today. Working on the rusticated ground floor of Sampson's Bank. Three different wall hosted families, moderately complex, with nested face-based void sweeps to get that chamfered stone block look. Dipping into my classical columns library also. A bit of attention to detail needed to get consistent alignments when turning those little corners. Classical details can be quite demanding, but also rewarding. Learnt a lot.



So far I have three different families using this approach.  None of them are really quite finished.  They tend to have a mix of the two approaches, with projecting string courses and keystones treated as veneers.  I will need to redo the families on the courtyard side also, and to sort out wrapping at all the reveals.  You can see the difference between the two approaches in the next image.



The face-based family itself has an instance length parameter and a sketch-based profile for the void sweep that does all the work.  Some day I may upgrade this to a loaded profile to allow for different joint styles.  I’m happy to have made some progress in representing this kind of masonry wall in Revit, but there’s a lot more to be done to really crack the case I think.



I added wooden doors for the three openings.  The side doors for pedestrians are straightforward enough, but the central archway needs to handle carriages so the door leaves have curved heads.  Given that these are thick walls and I have used groin vaults for all these external passages, based on buildings I know in London and thinking of security.  So when I decided to show the door leaves in an open position it became obvious that there was an issue with them snagging on the internal vaults. 

I have recessed the inner arch and pushed the vaults up higher.  Let’s try to face up to these practical issues if we can.  There are fireplaces and chimney stacks to develop further, internal paneling for the transfer offices on the upper floor.  Looks like the attic floor needs to be raised where it clashes with the central half-round window.



Lots of little details to add / resolve but I think it’s time to move on to something else and come back fresh to Sampson’s bank in the next round of updates.  I’m thinking it’s time to tackle some of Taylor’s work.  Probably start with his screen walls which I have never looked at very closely.



Sunday, December 27, 2020



Project Soane, extended version.  The Bank of England began as a group of non-conformist merchants,, many of them Huguenots, who came up with a scheme to elbow their way in to the mainstream financial world of 17th century London.  It was a variation of a proposal that had been tried before, but on this occasion, the timing was right.

They proposed to raise large amounts of money for the government to fight a war with France, by selling shares in a joint stock company.  This company went on to become the premier financial institution mediating between government and private enterprise: the Bank of England. 

They began in a small way in rented premises belonging to the Worshipful Company of Grocers, set back from the main street in an enclosed courtyard.  I started to model this building a couple of years ago but didn’t get very far and it seemed like a manageable task to occupy me for a few days taking it to a shareable level.


The family I originally made for the pointed-arch windows came before Project Notre Dame and wasn’t really parametric.  I decided this was a good time to revisit my modular door/window families and extend the system to cover different kinds of arched heads.  I also wanted a standardised approach to adapting the system for archways with no nested components visible in the project.

So now I have four variants: half round, flat, pointed and segmental.  The formula requirements for these are slightly different, but once set up they work consistently and different nested components for the frames and “doorsets” can be swapped out very easily.  Double, single, plain, panelled … frames with fanlights.  It’s a good start.


Back in the Grocer’s Hall things are shaping up.  The large triple-height central space was the great public pay hall where all the major financial transactions took place.  I don’t have much information on this, (no interior views) so it was an interesting opportunity to do some conjectural modeling, trying to conjure up the atmosphere of a medieval livery hall.


The main source of information here is Daniel Abramson’s book “Building the Bank of England” which has some hand-drawn floor plans (probably deduced from written descriptions?) and a front elevation showing the main façade and its forecourt, date unknown, I’m guessing around 1700?  One of my main goals is to understand and communicate the organisational structure of the bank in it’s earlieast incarnation.  This will help us to track how that structure was mapped onto new spaces as the bank continued to expand.

So I have managed to develop this model far enough to set up two drawing sheets with a variety of different views, some reference images and a bunch of text.  Some of it is loose text some has been embedded and tagged.  Ultimately it would be good to covert all of this into embedded data, but it’s not altogether clear where to embed some of this history. 

Be that as it may, I’m quite pleased with the way the story is conveyed.  I have linked back the context model from a previous post so that you can see the relationship of the Grocers Hall to the future site, and to the Mercer’s Hall which they rented briefly in 1694.  This illustrates very clearly how they moved closer and closer to the Royal Exchange as they grew in size and confidence.

You can download a 2 page PDF here

Grocers Hall Sheet Set


Sunday, December 13, 2020


Continuing with Sampson's Bank.  Some strange families in here.

  • 1.       Project Soane (PS) is more than 5 years old
  • 2.       The geometry is complex (Corinthian capitals, bulls head motifs)
  • 3.       It’s a large group of buildings
  • 4.       The reference material is full of gaps and difficult to interpret

So many of the families are placeholders, first attempts, quick sketches. I was feeling my way, improvising and taking shortcuts. Often there was little to be gained by trying to make them parametric, every chance they would break on flexing.

The Sampson’s Bank model has inherited families of different ages.  Usually I can’t remember what I did, so if they need to be updated I come face to face with a past version of myself.  I have to work with old and clunky families that predate my current modular systems. Most likely those systems would not exist, but for the "Project Soane Experience".

 But let's start with some new “half-round” columns for the entrance façade on Threadneedle Street.  These were removed by Soane around 1820, so they don’t feature in the PS Main Model.


Four columns, same basic design, two against a flat wall, two wrapping the corner.  I have my collection of classical columns, Planting category so they scale, and modular so they mix & match.  Base, Shaft & Capital.  Put these three elements together in whatever combination you need.  Economy of effort.

So I homed in on Ionic type 1, round, smooth, no pedestal.  40-Ionic-01-X-Smooth-RD.rfa  The X stands for “no base”  The types “full”, “deep”, “half”, “shallow” are built in to all these families by changing the depth of a void cut.  Actually there are void extrusions in each of the three sub-components that flex together to achieve the results.  Finally I have an instance parameter to convert to a “corner” (Yes/No tickbox) That works on the width of the void extrusion.


When I put this family in place and adjusted the height to match the top two stories, I realised that I had chosen the wrong one.  Easily fixed.  I renamed the family in the project browser, changing X to B.  Then loaded In the other family and let it over-ride.  The complex geometry of the Ionic Capital shrinks very slightly so that the total height remains the same.  The base was a bit too high so I adjusted this in family editor, bringing the shaft and capital down to suit.  Reload and the geometry enlarges very slightly to keep the overall height the same, as set in the hardwired type parameter.  The wonders of the “Planting Category Scaling Hack”

Looking at this in 3d I realised that the entablature moulding (an in place sweep above the columns) need to come forward a bit at the front and move in slightly at the sides.  Capitals always project beyond the beams they carry, but the proportions matter, and alignments should be consistent throughout.  There is a slight problem with the string course below also, but I didn’t fuss with that just now.  Very minor and lots of more pressing issues to tackle.  Actually, if you look carefully you will see that the corner of the wall is pushing through the shaft of the column and the window is not quite centred between the columns.

Classical architecture is a wondrous system but very demanding.  I enjoy the discipline, attention to detail, deliberate proportions.


At the back of this block, facing the internal courtyard, the façade was not modified by Soane.  The ground floor arches were glazed, probably by Taylor, but otherwise the elevation in the PS Main Model remains as designed by Sampson, so I was able to copy-paste this virtually intact.  Here the columns are square and the families used predate my modular system.  I may replace these eventually, but it’s not urgent.

Interesting to open this up.  An extrusion and a couple of sweeps, placed directly in a wall-hosted template.  The volutes are two cylindrical extrusions placed in a nested family because they are at 45 degrees and they need to be locked back to the wall face.  I’m sure I knocked this family up one afternoon quickly, probably with only a couple of old photos for reference.  It is possible to achieve a passable placeholder for the classical orders using simple shapes.


In the Pay Hall we also have engaged columns of the Ionic Order, but this time they have full-blown plinths rather than a simple square base.  This one is quite interesting when you open up the family.  It’s derived from my modular system, but with a twist in the tail.  It’s a wall hosted template with a nested family.  The nested family contains a “free form element” That means I chose the family I wanted, sized it to the room, exported to acis.sat (solid CAD) then brought that back into a family template and exploded it.  At the time I did it this resulted in an object with lots of shape handles that will take a material parameter.  That behaviour changed a couple of releases ago. 

Why did I do this?  I must have been concerned about the levels of nesting.  The capital has a couple of nested components.  This is then nested into a family called “inner” which itself gets nested into the column as found in my collection.  Two levels of planting family (double nested planting) in order to get the scaling behaviour.  To make this wall hosted would mean yet another level of nesting which starts to push the file size up and can slow things down quite a bit.  That would have been my reasoning at the time.



Getting back to the 4 new columns. To fit on the ledge available, I had to make “ a small half”.  This changes the proportions a bit, and the shafts of the corner columns will look slightly wider than the others.  I think I’ve got it about right, but I will have to look more carefully at real-life examples.

I will have to work on the flat pilasters.  Not so much concerned about the simplicity of the capitals as the need for a corner version. I’m not sure how this was how Sampson built it, conflicting evidence and Taylor seems to have built scrolls up against those corners later on.  I think corner pilasters is the neatest solution for our study model.  I’m also going to express the chimney breasts on that end elevation to give the cornice mouldings something to stop against.