Trying something new: A 1:6 scale miniature set, with 3D tracking and digital extensions…

Having worked exclusively in digital formats for the past 15 years, I have recently been making a concerted effort to resurrect some of my practical design skills. Around five years ago, I started to sketch more frequently, as I began to take on more of a design lead on projects. It proved a great complement to pixels.

More recently, I have looked to expand upon this positive experience and started to explore the use of miniatures in filmmaking. The shots below are of a miniature I have been developing for a sequence of shots based on a snowy environment I have designed. Two faces of a section of a building have been constructed, and set within a snow covered landscape. This miniature set design will ultimately be extended through the use of digital set extensions and 3D tracking.

Shot of the in-progress, 1:6 scale, miniature set, featuring snow and ice effects
A shot of the miniature including snow and ice effects.

The shell of the building was made out of foamcore, which was then clad in balsa. The balsa was ridiculously pristine when first applied, but has since been aged and weathered with heavily diluted ink washes. The snow is a combination of polystyrene base, on top of which, model railway snow effects and marble dust (to add sparkle) have been applied. The icicles are made by applying model railway water effect gel to waxed paper. Once dried, they are glued to the eaves of the building. It’s not finished yet, but so far, it’s proved to be a hugely, enjoyable indulgence.

Detail shot of 1:6 scale miniature set, featuring snow and ice effects
Detail of the miniature in it’s current stage of development.

As with any miniature, depth of field has proved problematic. The only way to counter this, has been to dramatically stop down. Testing has shown f16 to be the absolute maximum, but to achieve optimum results, I am going to need to reduce the aperture even further. The direct consequence of this, is that I then need ridiculous levels of light.

I’m currently looking at stop motion techniques to achieve small apertures, with less light and longer exposures. This approach will allow for better digital environments through better depth of field and improved image quality and resolution compared to video capture.

I wanted to post these before Fortitude is screened tomorrow, as I started work on this last Autumn but business has been very busy and limited the opportunities to work on these kinds of personal projects.

Work schedules permitting, I’m hoping to have the first sequence completed by the Spring

An earlier shot of the miniature set, prior to adding snow and ice effects.
Shot of the miniature prior to the application of snow and ice effects.


Sanchez Law – “It’s easier to do it right, than it is to do it wrong”

When I first started out in architecture, CAD was a thing, but when it came to drawings, manual draughting ruled. The first time I encountered CAD, was whilst working in Hong Kong, for a company called ArcPac. There, they had four workstations running AutoCAD 11. Recognising the potential, I quickly made myself useful on these machines and began to learn the dark art of computer aided drawing within a multinational team. We had a big group of Filipinos working in the office at the time, among them, one Jocelyn Sanchez (Jos). Jos, was brilliant; dedicated in her work, she never suffered fools, but was always patient and helpful toward those prepared to work hard, and give things a go. The CAD skills that I learned in Hong Kong, proved the foundation for the 3D skills I would later use in my TV work. To this day, I still repeat the mantra that I learned from Jos, namely that when using computers, “It’s easier to do it right, than it is to do it wrong”. We came to refer to this as Sanchez Law.

Jos was referring to the ability of a package like AutoCAD to produce drawings of previously unparalleled accuracy. If you wanted to draw a line exactly 1200mm long, then you drew a line 1200mm long. What used to wind Jos up, would be working on someone elses drawings, and finding lines that were 1200.3mm long, instead of 1200mm. Not only did that .3mm mess up the dimensions, but like the pea beneath the mattress in the fairy tale, it’s inaccuracy would infect every other part of the drawing. What made this worse, was that the inaccuracy would inevitably be the result of sloppiness. Indeed, it really was harder to make a line 1200.3mm long, than it was to make one 1200mm long.

I don’t use CAD much these days, but I do of course make plenty of use of 3D software, namely 3DS Max. I still work accurately, and to the mm as far as practically possible, but when you have the ability to input using parametric modelling, why wouldn’t you? If you need an object to be a specific size, then you create it at that specific size. Buildings and sets are, for the most part, pretty rectangular and ordered affairs. Curves might be a bit trickier, but it’s still easy enough to create them with sufficient precision. If a rectangular, or even a circular element is meant to be 1200mm across, create it 1200mm across…unless of course, you don’t know that it’s meant to be 1200mm across. Maybe that’s the problem? Are people producing inaccurate work, because they don’t fully understand what they are describing in their models and drawings?

I do sometimes get sent other people’s models, but have learnt to refuse to use them, except for reference. When you take care with your own work, it’s infuriating to find yourself within someone elses model, where the ground plane should be at zero, but is in fact at 104.128mm. Life (and work) is so much simpler if we do things properly, when we are provided the tools to do so. Because the ground plane is at 104.128, nothing else will be in the correct place, and every derivative part of the model will be equally inaccurate. I don’t understand how 0.00 ended up as 104.128, because my experience (and Sanchez Law) tells me that it would have been easier to do it right, than it was to do it wrong.

Jocelyn Sanchez, you were a genius, ahead of your time, and I’m glad that my 23 year old self listened to you back then!

Dredd 3D – Building Blocks in need of a few real world constraints


A short, behind the scenes piece on the design and VFX work done by Prime Focus to create Mega City One for Dredd 3D. The filmmakers chose to build their fictional city by adding to the existing fabric of Johannesburg. The realism appears grimier, lived in and more tangible as a result.

If I have one reservation about the designs in these digital set extensions, it would be that the blocks are treated purely as forms without reference to their function or environment. Specifically, the floor plates on the blocks are really massive, which would make the cores deep within these buildings oppressively dark in a way that would make Hong Kong’s Chung King Mansion seem generously light and airy in comparison. Even a slum needs light, and the struggle to introduce this light would lead to a different footprint and therefore an alternate form for the towers – time spent studying architecture does have its benefits.

As much as I like the look of these visuals, I would like to have seen the design extended further through the application of some real world constraints.

From the archives…

The process of redesigning and re-building the lightwell website is proving to be a surprising trip down memory lane. In over ten years of designing and illustrating architecture and buildings I have amassed a huge portfolio of images, many of which get archived and forgotten after a few years. A lot of the images get passed over because technically and artistically they get superseded by better examples. Others, I just get bored of looking at.

Architects Journal Front Cover, May 2000

The images I created for Luz Vargas Concept House 2000 designs probably fit into both those categories now, but for a time they proved invaluable to me in starting out as a freelancer as Luz was managing to get them printed and showcased everywhere. As well being exhibited in the Royal Academy summer exhibition, they managed to make the front cover of the Architects Journal.

If nothing else, I think this particular project demonstrated perfectly the benefits of having an enthusiastic client with an interesting design to work on when illustrating or visualising architecture.

3D Forests

As an artist, every now and then you need to pull your socks up and catch up not only with the competition, but also the tools you use each day. I had one of these mini-epiphanies late last year after winning a comission to prepare a set of images for a new development set in forested hills on the island of Grenada. Usually, when forests and planting are required I would prepare the buildings and terrain using 3D but then add the trees and planting in Photoshop using photomontage and digital paint…lots of it. On this project I wanted to achieve something new and so opted to gamble on an all CG landscape approach.

Despite having tried various CG solutions to landscaping in the past, I found that either render times became prohibitive or the solution I was hoping for just wouldn’t work for me as advertised – Vue XStream comes to mind. As far back as 2003, I bought the Onyx Tree/Storm plugins for Max – I’d been amazed by a Tree Storm animation of some palms blowing  in the wind rendered using Electric Image at Londons DMW in 1997 and could never quite let go of that imagery – but the hardware and rendering limitations of that time meant that, as sophisticated as its trees looked, they just weren’t viable for a project with a deadline and the software remained uninstalled and pretty much overlooked.

River View Detail

What’s changed? A few things. Apart from the availability of more powerful hardware and software, I would cite  Peter Guthrie’s work using Onyx Tree as a motivational spur. Around the same time, I had belatedly discovered the performance benefits of using Vray Proxy meshes. Finally, I was pointed in the direction of a fantastic Max plug-in from Itoo Software called Forest Pack Pro.

Forest Pack Pro allowed me to populate irregular terrain with Vray Proxy models of tree and plant meshes (or rocks in the case of the river bed) with relative ease and because it references these meshes rather duplicates them, it keeps memory usage incredibly low and render times fast and manageable. The whole scene (shown in the aerial shot above) amounted to just 1,650,843 polys.

Aerial View Detail

I could have painted all of those trees digitally in photoshop, but given the number and variety of images required on this project I don’t feel I could have done it as well or as efficiently. Besides, I quite enjoyed playing and scattering trees about the landscape without having to worry to much about Max keeling over.