Wednesday, May 31, 2017


I'm heading for England tomorrow night, visiting my grandson, playing some music, calling in at the Soane Museum.  Looking forward to a change of scene for a couple of weeks.

Before I set off, I wanted to report on the work we have done modelling John Soane's Board of Trade Building.  There's still a fair amount of work to do, but it's coming along nicely.  Quite a decent sheet set coming out and lots of interesting questions about the design process have been revealed.

You can see a detailed report on by following this link, which itself has links to a downloadable sheet set and three panoramas exported from Enscape 3d.  I have to say that the quality and variety of images you can generate in a short space of time with this plugin is really impressive

Thursday, May 18, 2017


I had a problem syncing to C4R last weekend. Checking the cache I realised that the file size had jumped up rather alarmingly.  It's the Site Context file for the Board of Trade building on Project Soane, and I had been working up some design development studies using Phasing.  At first I thought phasing might be the problem (can definitely slow things down) but the way I fixed it was by simplifying my Corinthian Column family.

It's a massing study so I don't need all that detail anyway.  But next morning I was struck by the idea of using a new approach to a simplified Corinthian Capital.  Basically it uses two octagonal blends, (one rotated 22.5 degrees in relation to the other) to stand as placeholders for the two rows of acanthus leaves. For the scrolls I used extremely simple extrusions and the whole thing turned out to be extremely lightweight: 600kb as opposed to the 6mb of my original.  I had been quite proud of getting a Corinthian column down to 6mb.  One of the versions we had during the competition stage of Project Soane was over 100mb.  But there's a lot to be said for a really lightweight placeholder.

By elaborating the sketch for the blends slightly I got something even better with a negligible increase in file size.  But then I got to wondering what it is that pushes the file size up by a factor of 10.

I tried deleting different parts of the capital one by one and soon discovered that my scrolls were the main offender.  So although I had started this exploration with an idea for replacing the leaves, the real breakthrough came from remaking the scrolls.  And in my view they were the weakest link anyway.  I was never really happy with the way they looked, but hadn't found the energy to try again.

It seemed to me that the trick was to just use a series of solid extrusions.  Has to be the lightest of lightweight modelling methods.  So I made a sketch like an apostrophe (or a tadpole) and to be more faithful to @theRealCorinthian, had two of these at each corner, splayed apart slightly. Then borrowing a trick from the Funny Fireplace post I did a couple of weeks ago, I added a cylinder that passes through both and projects slightly at each side. 

I'm happy with this as a way of tricking the eye into seeing what it expects to see.  The result is a Corinthian Column that weights in at about a quarter of the size of my previous attempt, and to my eye it looks a good deal better also.

I get a lot of pleasure from this kind of exercise.  The art of "keeping it simple" is very satisfying once you hit on the right approach.  So that's my current best effort at the Corinthian Order: a double-nested planting object (and therefore scalable) that weighs in at around 1.5mb 
I will be updating my column collection and re-sharing it shortly.

Tuesday, May 2, 2017


Project Soane continues to surprise me with new twists and turns.  The Slack group we started a few weeks ago has proved very fertile.  I do encourage more people to join (or rejoin) our little endeavour.  If you have any interest at all in History, Buildings, Architecture, the Enlightenment, the Industrial Revolution ... this is a splendid way to pursue that interest in a very practical way.

We have been exploring the broader sweep of John Soane's career, and in particular his work in and around Whitehall, an area of Westminster which has become synonymous with the executive arm of government in the UK. He built Law Courts, nestled into the ancient buttresses of Westminster Hall, produced numerous schemes for alterations to the old Houses of Parliament, repaired and maintained the Banqueting House (Inigo Jones) and the Old Treasury (William Kent)

The Law Courts project is fascinating, but was demolished in Victorian times.  Even shorter lived was his Board of Trade building, (although parts of this remain).  This is the focus of our current work and has entailed some revisiting of the Timber Sash Window, which I studied some years ago


That study was a non-parametric analysis of how the technology works, and I used Revit because it's my favourite drawing tool.  The ease with which you can develop multiple orthographic views of an object from a 3d model, (dimensioned, annotated and arranged on sheets) is hugely valuable when you are exploring building technologies, whether old or new.  I am dumbfounded that this has not caught on more as a didactic method in Universities around the world.  

Over the past couple of weeks, I have taken the knowledge gained in my previous study and applied it to creating useable families for our model of the Board of Trade building, at the corner of Whitehall and Downing Street.  There are two models, loaded up into the Project Soane website on BIM360 and accessible using C4R.  The first is a context model, quick massing studies of some of the very interesting historical buildings that are crammed into this area, so familiar to tourists from all around the world. 

In the view below, they are colour coded by age.  Those that were there in Soane's day are solid. Those built later are transparent.  Yellow means medieval. Orange is 17 century and Red 18th. Soane's Board of Trade is the more heavily detailed, monochrome, L-shaped block. 19th century is transparent pink, and 20th very pale violet.  Westminster Bridge was the second bridge to be built across the Thames, and was still quite new in Soane's day.

You can immediately see that the Board of Trade represents the beginnings of a huge expansion of the Civil Service that was to occur during the Industrial Revolution as Britain flooded the world with Cotton and other factory-made goods.  This was the first model I ever opened up in Enscape and I was immediately impressed by this shot of Whitehall from the Thames, with Westminster bridge and the Houses of Parliament on the left, and the 1940s Ministry of Defence, centre right.

At that stage the Board of Trade itself was fairly crudely modelled.  This is the second file on C4R of course, and here it is, viewed using Enscape, with Downing Street centre left, heading towards St James Park. The space behind the buildings (top right) is Horseguards Parade, where you can watch the "changing of the guards" ceremony: uniformed men with swords, on horseback. 

It's a remnant of the age that was slipping away almost imperceptibly during Soane's lifetime, when the Privy Council Chamber was still a meeting place where senior members of the aristocracy met to advise the king.  That's the open-topped space that terminates the shorter leg of Soane's building as it projects into Downing Street.  By the way the building is incorrectly labelled as "Soane's Treasury" in these images.  I was taking my lead from Wikipedia which is quite misleading. It is now known as Treasury Buildings, and houses the Cabinet Office but in Soane's day housed the Board of Trade and Privy Council Offices.

The buildings in the Site Context model are represented by families. This keeps everything nice and lightweight, while allowing properties to be added and scheduled, so we can keep track of the architects and dates of construction.  Eventually I would love to apply phasing to this model and create some kind of a time lapse study.  There is also a sheet of compiled web clippings recording research done to date.

Getting back to the windows, I have been pursuing "modularity" for some time in my families, using reloadable, nested components, often with standardised names so that all the parameter linking is preserved when you load a different version.  This is quite fully developed for doors, and I'm in process of extending it to Windows. The nested components are "SASH" and "GRILLE" (grille is nested inside sash) plus, more recently, VOID (which is a replacement for the usual opening cut)

Often these windows have splayed internal reveals that extend to floor level and may contain folding shutters.  By making the void cut as a nested component, we can swap out different versions.  Later sash windows fitted into thinner walls would be finished more simply with architraves and a horizontal wooden sill, finishing flush with the interior wall face.

I haven't modelled any internal shutters yet, but the splayed reveals are showing up nicely.  I love the quality of light in Enscape's internal views.  It takes so long to get a decent internal render with Revit's built-in engine, but here we are getting splendid results almost instantly. Notice that the 'trial version" watermark has disappeared now.  I have activated the 6 month license that they so kindly donated to the cause of Project Soane.

The glazing bar profiles started as a simple rectangle, but later I developed a parametric version of the typical "lamb's tongue design" that became more or less standard.  It's amazing how much slimmer they appear, after adding a couple of delicate mouldings to the corners of a square piece of timber.

There are a few basic formulas to allow for equal-sized top and bottom lights (or not), with a simple tick box.  I had more elaborate versions that allowed the setting of different glazing bar configurations parametrically, but this started to affect performance, so currently I have a collection of "GRILLE" components to cover the different cases, and load whatever I need for each particular case.

While I was at it, I spent a day or so building standard steel windows on the same lines. These are very common in Africa and provide an excellent solution, given the local climate and the nature of the construction industry.  I developed a set of Autocad blocks for these over 20 years ago, so it felt quite good to do the "BIM equivalent"  I hope to share these later on, once I've cleaned them up a bit.

I also did a couple of posts on steel windows, as a follow on to the sliding sash study.  You can take a look    HERE

Meanwhile the timber windows are looking quite good in a plan view.  The inset shows the state of the model at this stage.  I decided to break off the family development exercise and put some time into the building itself.

A day or so later the rear elevation is shaping up nicely.  It seems that Soane incorporated an existing house into his design (the previous location of the Board of Trade, I think)  At first his commission was simply to enlarge this house, but at some point the Privy Council Offices were added to the brief and the project doubled in size.  I love to trace his design development processes and building a Revit model is a great way to query the source material. To one side of the original building he added (or extended?) a staircase which provided access to the new attic rooms along the Whitehall frontage.
There is a magnificent central corridor with barrel-vaulted skylights, running down the centre of the main block.  Figuring out how this works was another rewarding detective puzzle.  As usual he is quite inventive in the way he divides the wall surface up with decorative panels.

Anyone familiar with London will know about the recessed "areas" between pavement and building that provide light and ventilation to the basements.  Very often these are semi-basements, two thirds of a storey down from the street, with a substantial number of stairs leading up to the ground floor entrance doors.

I always think that one of the marks of a good architect is the ability to blur the lines between spatial drama and pragmatic solution.  The central corridor is/was a magnificent space but it also serves to bring borrowed light into the attic rooms, separated as they are from the external wall along Whitehall by an access passage.

I have a tendency to flit back and forth when I'm modelling.  I suppose I'm trying to balance things off, gradually raise the level of development across the whole model, whether that should be "getting the planning right" or "fleshing out the details"  Nothing quite like an entablature moulding for raising the impression of completeness.

Mr Soane did like his complex roofscapes, but there's always a rationale in terms of bringing light in from unexpected angles, especially into stairs and corridors, which can so easily become such dismal spaces.

Starting to develop sheets in the model now.  It's interesting to compare the front and rear elevations.  He has created a grand colonnade along Whitehall.  In fact he produced many, many schemes to extend this further, but was thwarted by their lordships, (his clients).  But to the rear, the impression is of a row of varied terraced houses, very much the way that the Whitehall façade was before his intervention.  So he leaves a memory of the past at the rear of the building.  Note how I'm still using placeholder families here to represent the windows.

Adding rooms (with the automatic placement feature) allows us to use a colour scheme to show how the building is divided into two departments. Hopefully we can allocate names to most of these rooms by poring over the drawings and accompanying notes in the Soane Museum online archive.  I've said it before, but what an incredible resource!

And finally sections through the main spine, looking both ways.  You can see the borrowed lights at high level, and the way he uses the moulded panels to hide the irregularity of their placement.  Lots of detail to add yet, there are brackets at ceiling level (modillions) and Greek key designs in the long horizontal panels.  What larks Pip!

Wednesday, April 26, 2017


I recently received a message from Robin Deurloo, who lives in the Netherlands and is one of those who made an effort to contribute to the modelling stage of Project Soane. He had received a request to model a fireplace design which is in a vaguely classical style and was struggling a bit with one element.  He thought of me, and reached out.

I guess lots of people who find themselves "modelling" on the basis of someone else's "design" face these kind of issues.  The sketches he was working from are somewhat tentative/ambiguous.  That's in the nature of quick design sketches really.  In this case there are inherent contradictions between the different orthographic views, and quite frankly, it suggests a fairly shallow understanding of classical principles... But the general intentions are clear enough, and both Robin and his client have made genuine requests for help, so I decided to have a go.

The problem area is the scroll.  My own feeling is that you would never use a scroll quite like this. It seems somehow incomplete and disconnected from the rest of the design. Scrolls tend to follow and "S" shape with one side larger than the other, and assume the function of brackets (at least symbolically)  Also the Scroll as drawn is something of an Escheresque impossible object. The projecting portion and its background merge into each other as you follow the curve around. 

So my own free interpretation of the intentions are shown above.  I'm ashamed to admit that I had never drawn a spiral by the "proper method" before.  I've always fudged it in my impatience.  Considering that I wrote the foreword to Paul Aubin's excellent book "Renaissance Revit" some 3 years ago, this is quite an admission.  So I reached over for Paul's book and constructed a rig of centre points that gradually work their way around, moving inwards as they go.  The scroll as shown in the sketch didn't have so many turns, so I adapted the results with a larger centre circle.

That gave me a filled region which I can edit, and copy the sketch into another context.  You can use the sketch from a filled region to make an extrusion by copying from sketch mode and pasting into sketch mode. And you can select all the lines in a sketch and scale them up and down to your hearts content as long as none of the segments become too short.

Now it is fairly common for spirals to project further out as they wind in towards the centre, and the sketch provided implies this.  I have ignored this for the moment.  You would need to create a swept blend, probably using a spline that approximates the curve of the spiral.  I'm not quite convinced that the relative widths of the various components has been properly thought through in this design, so I decided to keep things simple, at least until the proportions have been resolved.

For the same reasons, the fluting on the extrusion below the scroll is not curved in cross-section.  The whole composition is rather squared off and crude, but that's OK.  I am simply presenting an analysis for discussion.  This is as close as I can get to the design sketch while maintaining a coherent 3-dimensional language.  From here you could go on to discuss further adaptations and embellishments.  Personally I would be inclined to either accept this solution, or rethink the whole thing rather radically, but that's just me.

You can download my little family HERE, and if you look carefully, you'll see an embedded Detail Item that illustrates the setting out of the spiral.  This is not 100% as Paul's meticulously explained method.  I lost patience when it came to the inner spiral and reverse-engineered an approximation, but it's close, and you can scale this to a variety of applications with reasonable success.
I have interpreted this scroll as half an Ionic Capital (rather than an incomplete and upside down bracket) but there many other ways of using scrolls.  The classical language of architecture is remarkably flexible.

Personally I am a little squeamish about designing modern buildings in a classical style: it rarely comes off in my view.  But I get a huge amount of pleasure from researching buildings of the past using Revit, and if nothing else, this exercise has proved to me once more how remarkable Paul Aubin's book is, as a hands-on learning experience.  If you have it, give it another shot.  You probably found it hard work first time around, but it really repays the effort of repeated study.  If you missed it first time around, then go and grab a copy.  Fantastic value

Wednesday, April 19, 2017


I didn't really get any for my birthday (toys), but I have been playing with a couple of add ins, and of course the new features of Revit2018 have been announced. (no comment)

During BiLT Asia, I met lots of interesting people, including founding members of Enscape3d and I've had my eye on both of these initiatives for some time now, but talking to people intimately involved with the development of a project always helps to push it towards the top of the priority stack.

Thomas Schander from Enscape 3D was kind enough to offer me a six month license to use with Project Soane, having loaded up the model on the system at their booth with quite impressive results.  I haven't got very far yet, indeed I haven't even got around to registering my license, but I think it's going to be very, very useful.

The first shot is a two-point perspective of the Doric Vestibule, looking East towards the loggia which is highlighted by light streaming in from the Waiting Room Court on the right hand side.  I've turned the fog up, so you can see the shafts of light streaming in from the half-round windows at high level.

Next up is a view looking back at that open-sided loggia from at the South-West corner of Waiting Room Court. The virtual dust particles are still hanging thick in the air, so although you can see straight through the accountants office to Printing Court beyond, the effect is not very striking.  When we first tried the model out on the Enscape Stand we got a view from the Governor's Office window with the Printing Court positively glowing in the distance.  So far I have failed in my attempts to recreate this effect, but it's early days.

By accident, I slipped through the wall to the right, through some waiting rooms and into the Central Lobby of the Directors Parlours, an area that I worked on quite intensively towards the end of 2016 and beginning of 2017. This wide-angle shot gives quite a nice impression of the space with low, narrow corridor coming in from the Entrance Lobby, the high level lighting and central groin vault.

I also exported a panorama from this space. I downloaded a freeware viewer called Panorado which works fine, but I couldn't figure out how to embed it in a blog post.  Maybe there's a way to upload these to a gallery.

OK, I knew this really ... lots of links to 360Cities in Google Earth and it's easy enough to set up an account.  So here are some links:

And then of course I went back to Enscape and realised it's got a built in Gallery function, which is much easier.  But it expires with the license, so 360Cities is also useful to have.  Here's a link to Enscape, it won't last long though because it was created with a 14 day trial.

And then a couple of days later I realised that Google Photos has a built-in conversion feature, I had uploaded one of the panoramas to Google Drive, hoping to see it as a panorama.  Doesn't quite work like that, you have to go to the Assistant which then redirects you to a web link.  But you do get a message at some point telling you that your panoramas have been processed.

It's early days yet, but Enscape is definitely giving me a much deeper understanding of the quality of space and light in Soane's masterwork.  That's quite surprising, given how well I thought I new the model.  I guess virtual reality trumps an architects normal ability to visualise what it's going to be like in a space.  And for sure it will help in setting priorities as we raise the level of detail across the model.

So what about Flux?  Well all I've done so far is play around with the Site Extractor. Getting data directly into Revit doesn't work, you need to go via Dynamo, or in my case I used AutoCAD (oh no!)  I started with Whitehall (current Project Soane work) then tried a couple more flattish places (Robie House, Lever House) before realising I needed a hill.  Le Corbusier's chapel at Ronchamp is on a hill, so I grabbed that, then I thought about the work I did on Palladio's Villas back in 2014.

How time flies.  Did a couple of posts about this just over a year ago.

So I grabbed the site for Villa Barbaro which is built on rising ground leading up towards the foothills of the alps.  By then I had set up a starter file with appropriate layer presets for the data keys that Site Extractor grabs for you from open source maps. I've dispensed with using the topography surface because I couldn't get that to generate a topo surface in Revit.  Contour lines work fine, so no need anyway.

I created a CAD links file where I can generate topo surfaces in the right place without contaminating the main file with CAD nonsense.  Copy from there and Paste Aligned to Same Place.

I had to copy over some sub regions before deleting the original topo surface (that I built rather laboriously, estimating heights from Google Earth.)  It was way out of course, but that's all part of the extended learning curve (approaching 3 years now, since I started exploring Palladio's bag of urban design tricks.)

The tree families had been set up to host themselves on surfaces, so they auto adjust, but the buildings and floors needed manual tweaking. I opened this post with a shot of Behzad Meskini who has been doing lots of really interesting things with Revit (and Inventor).  I attended his session at BiLT Asia and among other things, gained renewed enthusiasm for the "Modify Sub Elements" tool for floors.

Nothing very new here, but I had overlooked the ability to grab horizontal lines and stretch them up and down.  This makes sloping roads (ramps) a lot easier to handle.  No variable layers checked in the floor type so the whole floor slopes (rather than getting thicker)

One thing that had put me off the Sub Element tool was all those diagonal lines that show up when surfaces get slightly twisted.  It's nice when you want it, but in this case I don't.  Easy enough to turn them off (and you can set up a View Template of course.)  So that was Villa Barbaro moved forward significantly with a couple of hours work.

This villa has a natural spring at the back which has been built in to a terrace up at the main floor level.  Ground floor facing the road is next level down.  It all makes sense now.  One day I will model this villa properly, including the stairs that rise up from the side wings up to the main level.  Already looking good though.

The whole study model is pretty crazy, over 100km across, but it's also very lightweight and agile.  Largely because all the villas are created as families.  Originally I had started doing the studies side by side in a rectangular space, but I was getting confused about where they all were on the map. So I moved them around so they were in roughly the right relative direction.  Pretty soon the ones close to Vicenza started moving to the correct distance apart so they didn't overlap.  Then it was a short step to putting them all in roughly the right place.

So I decided to do one more. 

Villa Rocca Pisani isn't actually by Palladio, but Vincenzo Scamozzi inherited Palladio's mantle and completed / extended several of his projects.  Also, it's dome is a probable source for England's best known "Palladian Villa": Chiswick House.  William Kent was involved in this project, one of the generation of English architects that preceded Soane and a notable contributor to the Whitehall area I have been looking at recently, along with my Project Soane collaborators.

I thought I had taken reasonably accurate heights from Google Earth, but it turns out my previous work was way out.  The hill rises at least twice as high above the adjacent plane (town of Lonigo)

In this case, I have a rather big topo surface representing the remnants of a ridge extending from the alps, down past Vicenza.  So I cut out a square hole in order to merge in the new topo.  We are not working to millimetre accuracy here.  Discrepancies of 10m are not going to be a big problem over of more than a kilometre.

The data coming is clearly approximate anyway, but it's much better than what I had before, and really helps in terms of what we are doing here.  It's a hands-on research project, aiming to uncover fresh insights into Palladio's work in context.

Click on the surface with the sub regions first when merging, or they will disappear. That's important!  But in this case I had to copy-paste the roads from old to new (in sketch mode)  Also there were points in my old surface, close to the edge of the rectangle, that resulted in crazy distortions.  Just delete these for a smoother effect.

I wanted to test adding a second set of data from an adjacent rectangle.  The villa is on a local high spot on ground that continues to rise from the plain, up into the mountains beyond. It worked fine, but aligning the two areas in 3d space is a manual task, which I did in Autocad.

Site Extractor just brings the data in relative to zero in Autocad each time.  No real-world position recorded.  That's fine.  Also the two sets of data don't match perfectly.  This stuff is fine for concept studies, don't expect it to work like survey data.

I included an overlap zone so as to make alignment easier to judge, so I had to trim this off before merging.  This is where clicking first on the topo with the sub regions comes into play.

Even after trimming tight along the edge there was quite a bit of turbulence.

All I did here was to delete points in a zone all along the edge strip, which disguises the discrepancy, smooths out the contours.

Here's the overall picture. 

Rising ground with a local high spot overlooking the town below in the river valley.

I was tempted to carry on adding more topo and working up the sheet, fleshing out the context, adjacent agricultural fields etc.  But this is not the time to return to my Palladio studies, I have to get back to Project Soane.  Flux site extractor works.  I've got my head around it, I know what to do next time I need to grab some topo, roads, rivers, approximate building footprints. 

All good, and thanks to Javier and Karl for taking the time to talk to me in Singapore.