A digital model of the entire World Heritage Site above and below ground
A very large project founded on a very simple idea – to capture with spatial fidelity in one digital model everything of significance above ground, below ground and underwater between Rangers House and Island Gardens.
That is, to build a digital model of the entire World Heritage Site incorporating the whole Royal Park above ground, the 300 odd tunnels and several reservoirs and conduits below ground, Rangers House, the Observatories, Flamsteed’s Well, Our Ladye Star of the Sea, Queen’s House, the Maritime Museum, the Tilt Yard, Admiral Hardy’s mausoleum above and below ground, the ORNC, St Alfege, St Alfege graveyard, the Tudor palace underground and what little is above ground, Cutty Sark, the foreshore, the Thames bathymetry underwater, the foot tunnel and Island Gardens. A half valley section cutting right through the World Heritage Site uniting every element of the architecture and landscape above and below ground.
The first model of any kind to unite and link every element of the World Heritage Site.
To make a digital model of what Canaletto saw when he painted Greenwich from Island Gardens in 1750.
Beer Quarry Caves sit about a mile inland from the village of Beer, on the Jurassic Coast in East Devon. They are not natural caves but the remains of a stone quarry worked underground for close to 2,000 years, from the Romans through to the 1920s. The stone itself, Beer stone, is a creamy-grey limestone that’s soft and easy to cut when first extracted but hardens with exposure to air, which made it a favourite for detailed church and cathedral work; it’s been used in Exeter Cathedral, Westminster Abbey, the Tower of London and Windsor Castle.
The Lixel K2 produces both a more conventional LiDAR point cloud and a Gaussian Splat representation of the environment.
The LiDAR scan
With no natural light – just low-level lighting strung through the cave network and a torch – the colour-capture element of the scan was significantly constrained. LiDAR, however, functions well in these conditions, accurately recording the location and position of elements, just without reliable true-colour information.
Substituting the actual, dark and muted colours with an elevation ramp – graduating from one colour to another according to elevation – brings the cave forms into clearer view and highlights changes in height throughout the cave network.
A custom OSL (Open Shading Language) file is used to define the elevation gradient, allowing more precise control over how colours are mapped across the vertical range of the cave. Applied through a V-Ray map in 3ds Max, this provides a consistent and easily adjustable elevation ramp across the entire point-cloud.
Since renders from angles viewing through the entire model can result in areas of overlaid and densely packed points, the view can look visually ‘blown out’, obscuring detail as the points compound and merge with depth. Several methods to remedy this:
VraySphereFade: to limit the render to just a sphere area – with a very soft (fade) cutoff edge
Render using a non-uniform aspect ratio to increase the Y-axis of the render area, vertically spreading the point cloud, reducing the overlap between points.
Set vray camera film gate very high – to dramatically increase the perspective and further separate the points
.
A contour calculation and visualisation tool, developed in-house, can provide an alternative, informative – and quick – method for illustrating the cave layout by slicing the point cloud and representing it as closed contour loops.
Contour Loop: ISO
Contour Loop: Plan
Contour Loop: Elevation
XGrids Lixel Studio enables interrogation of the point cloud through a sequence of 2D spatial profiles and the application of visual enhancements. It also provides volume calculations, including reference-plane volumes (the volume above or below a defined plane) as well as applications for earthworks and cut-and-fill analysis. These tools help present the cave as measurable data, rather than simply as a scan.
Lixel Studio
Lixel Studio
Lixel Studio
Gaussian Splat
The Lixel K2 also generates a Gaussian splat model of the scanned area – this is a representation of the scene as a collection of soft, volumetric elements, each with position, size, colour and opacity which produces a more continuous and visually realistic rendering.
Again, the limited lighting within the cave results in a relatively dark and visually subdued representation
A game-like walkthrough version is generated – albeit quite dark too.
This is a project to create a reproduction of a toki adze – a ceremonial axe from Mangaia in the Cook Islands – currently held in the Royal Museums Greenwich archive collection. The toki is approximately 850mm in length, with an intricately carved shaft of repeating cross-cut motifs, a polished dark stone blade, and fine coconut fibre lashing binding blade to haft.
The project sits within maker-identification research into Mangaian toki, undertaken with colleagues at Royal Museums Greenwich and connecting to work by carver Robbie Teremoana Atatoa.
Hundreds of Mangaian toki were removed from the Cook Islands during the nineteenth century and now exist in in museum collections around the world. While preserved physically, they are largely inaccessible to the communities and carving traditions they came from. This project explores whether modern photogrammetry and additive manufacturing can produce a faithful physical reproduction suitable for handling, study and cultural reconnection.
Photogrammetric Capture
This toki was captured over a supervised session in the RMG archive store, working with museum colleagues. Capture setup was tripod / DSLR-based photogrammetry using a Canon EOS R6 Mark III with a 24–105mm lens, shooting RAW at f/8–f/11 with manual focus and image stabilisation off.
The object sat on protective foam on a turntable, under LED studio lighting. White balance locked in to keep colour consistent across the full image set. The shot strategy was 36 frames per loop (every 10°) across four elevation passes; around 144+ frames for each side side.
Since the toki could not be stood upright – or readily handled – it needed to be photographed and flipped to capture each side separately. This breaks the geometric continuity that photogrammetry relies on, so the two sessions were essentially separate reconstructions from the outset rather than trying to force them into a single alignment.
Reconstruction and the Merge Problem
The image sets were processed with RealityCapture into dense meshes. The two flipped sessions reconstructed very well individually but as separate components – and with a slight scale discrepancy between them of roughly 1.1x – a known issue of turntable photogrammetry where the software has no absolute scale reference.
An attempt to resolve the registration externally in CloudCompare didn’t succeed, and the merge was instead worked through inside RealityCapture itself – placing manual control points across features visible in both sessions and iterating until the two halves pulled into agreement. The final alignment was close but not quite perfect – a fraction of residual mismatch that had to be accepted and tidied at the mesh stage rather than eliminated at source.
It’s a real reflection of flip-scan photogrammetry on an object like this: with no shared geometry between the two sessions beyond the object itself, there is a practical limit to how exactly two independent reconstructions can be forced together.
[note for future version: investigate small object laser scanner]
Limitations of Photogrammetric Reconstruction
Photogrammetry captured the overall form and the colour texture very well, but it has known limitations: sharp edges soften and deep dark crevices don’t reconstruct as they’re too much in the dark. On this toki, that meant the deep V-cut channels separating the 45 carved rings along the shaft came through too shallow – one of the most striking features of the object, lost just where the tool marks cut deepest.
To address this the geometry was brought into Blender 3d modelling software for some digital sculpting to reintroduce features visible on the original but missing from the mesh; sculpt mode sharpening to recover some edge definition and a series of torus rings Boolean subtracted from the shaft to recut those grooves.
This is the point it becomes clear this project has edged from replication / documentation into interpretation – an informed reconstruction rather than a pure scan copy.
Some sculpting back of the lashing zone on the head at this stage too – to make space on the finished print for real cordage, rather than a printed imitation of fibre binding which would never look authentic.
Fabrication
The reproduction was printed on a Bambu Lab H2D. At 850mm the toki far exceeds the 325mm build volume, so the handle was split into three sections in Blender, with the joints planned to fall where they could be disguised within the carved pattern.
PLA Wood Black Walnut and Rosewood – wood-filled filaments whose matt, slightly fibrous surface takes wood dye convincingly. The head stone was printed separately in a dark technical filament to read as stone against the timber tones of the shaft.
Print orientation took a test print to get right. The instinct was to print the handle sections lying down, but the results read badly: with the shaft on its side, the upper and lower surfaces of the cylinder meet the build plane at very acute angles, so each layer covers a wide sweep of the curve, and the stepping between layers becomes strongly evident – the carving dissolves into visible contour bands. Printed standing up, the geometry meets the layers far better: each ring builds as a series of near-horizontal cross-sections, and the V-cut channels keep more definition. A relatively coarse layer height (0.25–0.28mm) was retained deliberately; layering can work with a wooden object rather than against it.
Assembly
The three handle sections were joined permanently with cyanoacrylate adhesive – glue on one face, activator mist on the other, dry-fitted first, then held firm on contact. CA bonds well to PLA specifically, keying into the layer lines. Glue squeeze-out was kept clear of show surfaces, since cured CA resists dye and telegraphs through wax as a sheen difference.
Assembling before finishing was necessary as the shaft is dyed as one continuous piece, so the tonal gradient – richer and darker toward the head, lighter along the handle can run unbroken across the joints.
Finishing: Building Aged Wood in Layers
The finishing pipeline treats the print as if it were timber, sanding then using actual wood-finishing products rather than just paint effects.
An early lesson: water-based stain beads on smooth PLA. The answer was to work with dyes that penetrate the wood-filled surface – Liberon water-based wood dye as the foundation, with solvent-based spirit dye in a darker oak building depth. The first coat is applied diluted, which runs naturally into the recesses and settles pigment exactly where age would deposit it; subsequent coats build the overall tone.
[note for future version: solvent based dye to darken light coloured filament]
The worn, handled character of the original – lighter peaks where generations of hands have polished the carving – was reproduced by working a nearly-dry cloth of lightened tone (an ochre-tempered mix) across the peaks only, not into the grooves.
The final layer is Liberon Black Bison paste wax in Antique Pine – buffed hard on the peaks for the soft sheen of handled wood, and just lightly into the recesses with a toothbrush so the grooves keep their matt depth.
Lashing
The lashing is where the digital workflow stopped. The coconut fibre binding on the original is not a surface texture but a structural craft in its own right – a specific sequence of wraps and crossings that no 3D printing process can properly replicate. The binding on the reproduction was undertaken by Robbie Teremoana Atatoa, the only member of the project with the skills to do it correctly.
This means the reproduction is not wholly a machine translation of the original: its binding was made by hand, in the proper manner, by a carver whose practice descends directly from the tradition the toki belongs to. Where every other stage of the workflow is a copy of the object, the lashing is a continuation of it.
The finished toki is not the original. What it tests is how much of an object’s character – its wear, its tool marks, its accumulated handling – can survive translation through scan data and extruded filament. And unlike many of our reproductions, this one has somewhere to go: it travels with Robbie to New Zealand, a faithful, handleable copy within reach of the carving tradition it came from, while the original remains in the archive in Greenwich.
Credits
Scan photography undertaken with access to the original toki kindly facilitated by Royal Museums Greenwich: Ceremonial toki from Mangaia, Cook Islands (ZBA5519). Royal Museums Greenwich | Home
Project lead and cultural research: Robbie Atatoa. This recreation was undertaken in support of Robbie’s ongoing work to identify, translate, recreate and revitalise Mangaian toki, developed through his Churchill Fellowship research. Robbie Atatoa
Para Querer (No) Medir / For Want of (Not) Measuring is a collective exhibition that has been travelling internationally, initiated by artists Jim Hobbs and Patrick Adam Jones as an ongoing series, each iteration landing in a different city with different curators and invited collaborators. The Santiago edition was curated by Joselyne Contreras and Jorge Cabieses-Valdés, who brought together a group of Chilean and international artists at Sala Expo 1 of Centro Cultural CEINA – a cultural centre housed inside the historic Instituto Nacional building in the heart of the city.
The title frames the central question that runs through the work: what is it we actually do when we measure? And what do we choose not to measure – or choose not to let measuring dominate? For me, working with LiDAR, photogrammetry, Gaussian splatting, depth maps – as an expressive medium, these are significant questions. Precision instruments producing aesthetic rather than technical ends. Data becomes an image. Measurement becomes an interpretation.
I was exhibiting alongside Simon Withers, fellow member of Captivate Heritage Laboratory, along with Jim Hobbs (@jimhobbsart), Patrick Adam Jones (@patrickadam_jones), Andrés Durán (@andres_duran_d), Camila Lobos (@camilalobosd), Maria Brigita (@mariabrigita), Francisca Montes (@franciscamontes), Natalia Herrera (@nat.ssh), Jaime San Martín (@jaimesanmartina), and Jorge Cabieses-Valdés (@cabiesesvaldes). A genuinely collaborative, warm and supportive group – something that made the whole experience feel less like mounting a show and more like making something together.
The Work: Preparation (UK)
Element 1 : Very large laser scan render of sweet chesnut
The centrepiece of what was brought to the exhibition – a laser scan render of a 365-year-old sweet chestnut tree, a single tree, captured with LiDAR, processed and rendered as a point cloud image measuring 5,400 × 5,400 mm across six banner sections. Months of work in the UK preceded a single week in Santiago.
The preparation phase was substantial; scanning the tree being the starting point. Multiple LiDAR passes registered together into a single coherent point cloud and brought into 3ds Max / V-Ray for the points to be manipulated to read correctly – adjusting density and display, and most importantly selecting a viewpoint that could carry the weight of the subject. Throughout, resolution had to stay high enough so that when the image is tiled across six strips and printed at five metres each, it remains sharp and still holds as a surface worth looking at from close range.
Output phase: six banner sections produced on Canon Colourwave 700s, each panel a 35-minute print run. Two copies were made – one on 120gsm coated paper, which held the deep blacks the image needs, and a backup set on 130gsm Tyvek. The Tyvek is tear-resistant, useful for something travelling in luggage to the other side of the world, but it doesn’t give the same depth – the blacks come out smoky rather than inky, which matters when the image is built from points of light against darkness.
Element 2: Small hi-res renders of tree trunks
Alongside the large-format sweet chestnut banners, a set of four small, framed prints: tight, hyper-real scan renders of tree bark and trunk surfaces, printed at a scale that requires close inspection. The contrast was deliberate. A viewer would naturally walk up to the framed pieces – the detail pulling them in, the texture hyper-real at that range – then step left to find the full six-metre span of the tree demanding they step back to take it all in. The same subject, two entirely different registers of scale and body.
Element 3: (macro level) scans of seedlings / small plants.
A set of ten high-resolution scan renders – the macro end of the scalar range. Each derived from 100–150 macro photographs taken in studio conditions: controlled lighting, turntable, and in some cases a rotating hanging clamp. Processed via RealityScan, CloudCompare, and Blender with V-Ray. Blender’s geometry node system treats points as geometry – unlike 3ds Max – meaning size, shape, material transparency, and lighting all become variables in composing the final render.
Element 4: animation of point clouds
Alongside the still renders, a 30-minute looping animation of a separate set of seedling and plant scans – those that hadn’t come out well enough to present as stills. The animation is a screen recording of a semi-interactive web app in which each point cloud loads in a random orientation and applies random zooms, sweeps, and effects, with an optional sound-reactive mode. Running the live app unattended for two weeks in a gallery was too risky, so a recorded version was the practical solution – capturing the behaviour of the app without the vulnerability of leaving it live.
This work begins with measurement – a camera tracing the surface of a living plant through hundreds of overlapping photographs, reconstructing millions of coordinates in space from light and shadow alone. Rendered through Blender’s geometry node workflow and V-Ray’s physically based shading, each point in the cloud becomes a surface, a material, a thing that catches light. The process feels less like making and more like releasing – as though these images were always latent in the data, waiting to be let out. What emerges is less a record than a portrait – the plant known not through data, but through the attempt to capture it.
The Setup (Santiago)
All of this had to travel. Two sets of six five-metre banner sections, hanging hardware, print stock, hard drives – packed into hold luggage for the 14-hour flight from London to Santiago. One becomes very aware of the fragility of physical work at that point.
The gallery space at CEINA is a basement below a lecture hall with high walls on one side – genuinely high, which was the reason the tree banners were possible at all. Getting six-metre sections of print onto a wall without a proper rigging team requires patience, a certain amount of improvisation – and a very high ladder. The process of fixing, managing the weight and paper slip, getting the joints equal and level – it takes a day, and the work is cumulative. Fortunately, it arrived undamaged and stayed up om the wall so there was no need to swap to the reserve paper type after all.
Framing for the 10 gloss seedling prints needed to be sourced locally – some anxieties about navigating an unfamiliar city to find a frame supplier, but Santiago provided – the frames we found suited the work well and were remarkably cheap. A very ordered close 2 x 5 grid layout seemed fitting for this display wall.
The seedling animation played on a 40-inch screen on a standalone stand nearby close to the gallery entranced, a media player on a quiet loop. The animation shows each object from every angle, never settling.
Planning the layout alongside the other artists was a very rewarding part of the install. The exhibition as a whole had to function as a room – different bodies of work in conversation with each other, sight lines that made sense, space for viewers to move through. That negotiation, between artists who had never worked in the same room before, happened quickly and generously.
The opening was 12 noon on 23 May. The exhibition ran through to 25 June, open to the public Monday to Sunday, 11:00–20:00.
The Artists and the Room
What made the Santiago edition of For Want of (Not) Measuring work as a collective show was the variety of ways each artist approached the underlying questions of scale, precision, and what gets counted. The Chilean artists brought a different set of references and preoccupations from the international contributors — and the curators’ job, which Joselyne and Jorge did so well, was finding the threads between them without forcing resolution
The group was easy to work with. There was very much a mutual interest in each other’s practices, a willingness to help during install, and a low-ego collaborative atmosphere that made the mechanics of showing feel worthwhile.
Personal Reflections
Portraits in Points
This work begins with measurement – a camera tracing the surface of a living plant through hundreds of overlapping photographs, reconstructing millions of coordinates in space from light and shadow alone. Rendered through Blender’s geometry node workflow and V-Ray’s physically based shading, each point in the cloud becomes a surface, a material, a thing that catches light. The process feels less like making and more like releasing – as though these images were always latent in the data, waiting to be let out. What emerges is less a record than a portrait – the plant known not through data, but through the attempt to capture it.
Arriving in Santiago with the work – which I would always describe as a process, or the output of a process – it was a surprise (in a useful way) to see it received as “artworks”: discussed, interpreted, and placed in relation to other exhibition pieces. This reception shifted how I thought about the work, but the fact that I don’t consider it “art” is perhaps irrelevant. Designing workflows and systems, setting parameters, then waiting to see what comes out at the other end – whether that is art depends on the viewer, but the Santiago conversations raised the idea that this question itself is the important one.
One discussion that stays with me is around whether work needs explanation, or whether it should simply be put out there for the viewer to interpret blind. My instinct – certainly for this work – is that explanation matters. Scanning / processing / rendering is not something most people would be aware of, or aware has been used to produce what they’re looking at. One can look at these images and find them interesting or pleasing without knowing anything about their origin, but it’s the technology and the process behind them that makes them genuinely interesting to me. I’m aware I may be too close to this to be objective, but I feel the visual output is almost a byproduct of the process (the most important thing). I’m still not fully convinced that an uninformed viewing / reading is necessarily the richer one.
Another reflection is that there’s a sense in which this work is just cleverer than I am. The workflow produces things I couldn’t have designed or predicted. I set the conditions, build the system, choose the parameters – and then something comes out that surprises me. The intelligence, if that’s the right word, is distributed across the process rather than concentrated in any single decision I made. In a way, it just happened. That’s a curious thought, but perhaps the most honest description of how it feels.
What Santiago made clear was that work on its own – on a workstation screen – becomes something else when placed beside unfamiliar artists in an unfamiliar space. The role of the curator is not just organisation and layout, but interpretation – that’s key. The decisions Joselyne and Jorge made changed what the work meant – not in the sense of any loss of ownership, but by no longer being mine alone, it has another life.
Exhibition details Para Querer (No) Medir / For Want of (Not) Measuring Centro Cultural CEINA, Arturo Prat 33, Santiago 23 May – 25 June 2026 ceina.cl
This is a project to replicate 19th century African artefacts currently held by the Royal Greenwich Heritage Trust.
The workflow is: scan, 3d print and then authentically finish these historical items; an East African carved wooden headrest, a series of wooden hair combs, and bronze horseman sculptures. Together, these objects offer a range of material and surface qualities, from worn hardwood to aged metal, making them ideal for testing the full photogrammetry‑to‑copy workflow.
Rather than aiming to produce pristine, as-new replicas, this project focuses on how digital and fabrication processes can retain – and even amplify – traces of age, handling, and material character present in the original objects.
Photogrammetric Capture
All artefacts are captured using DSLR‑based photogrammetry. This approach allows for flexible image capture that adapts well to differing shapes and surface finishes.
Each object is photographed from all angles using a consistent lighting setup, ensuring strong image overlap and even exposure. Particular care taken with the bronze horseman, where reflective highlights can easily interfere with surface reconstruction. Diffuse lighting and careful camera positioning help minimise glare while still preserving subtle form and texture.
The carved wooden artefacts – the headrest and combs – are more straightforward, but the shallow relief and worn surfaces still require dense coverage to avoid soft or ambiguous geometry in the resulting models. In practice, capturing more images than strictly necessary consistently results in more reliable reconstructions.
Digital modelling and cleanup
The image sets are processed, via the photogrammetry software Reality Capture, into dense 3D meshes, followed by light digital cleanup in 3d editing software Blender. This stage is kept light – occlusion holes and obvious artefacts are addressed, but surface irregularities are largely preserved. This reflects an photogrammetry not as a means of achieving visual perfection, but as a way of retaining material history. Minor distortions introduced through capture can echo the wear already present in physical artefacts. Excessive smoothing risks replacing these qualities with an artificial digital uniformity.
Fabrication: choosing different printing processes
Each artefact here is fabricated using a different 3D printing technology, selected based on form, scale, and desired surface qualities.
The bronze horseman is printed on a nylon SLS (powder) printer. Due to its size, the model is produced in multiple parts and later reassembled. The slightly grainy surface typical of SLS prints becomes an advantage rather than a problem here, closely resembling the texture of cast metal.
The wooden combs are produced using a resin printer. This process builds objects by curing layers of liquid resin from the bottom up, resulting in extremely smooth surfaces once the supports are removed. This makes it particularly well‑suited to the fine teeth and delicate detailing of the combs.
The headrest is printed on a more conventional FDM printer, which extrudes molten plastic in layers, building the object up gradually. It is printed in two halves and then fixed together. The visible layering produced by this method echoes carved tool marks, making it a surprisingly appropriate choice for a wooden object.
Surface preparation
Before finishing, all prints undergo a combination of sanding, cleaning, and under‑coating. Preparation varies depending on material, but the intention is never to eliminate surface traces entirely.
Light sanding removes sharp artefacts while leaving enough texture to contribute to realism. Some pieces receive a single coat of filler primer, which is then sanded back so that it remains primarily in deeper print lines. This softens the surface without erasing its character.
Painting and finishing: simulating aged wood
The headrest and combs are finished using layered paint techniques designed to suggest aged hardwood rather than freshly finished timber.
A dark, warm brown base coat provides an initial foundation. From here, multiple diluted washes are applied, allowing pigment to settle into recesses and carved details. Raised areas are selectively blotted back, reducing the “plastic” appearance of the prints. Some judicious use of a nail to physically carve deeper into some grooves too.
Very restrained dry‑brushing is used to pick out edges and ridges. The emphasis here is subtlety – highlights that are obvious quickly undermine the illusion of age.
To finish, a neutral shoe polish is applied to areas that would naturally be handled. Buffed gently, this creates a soft, uneven sheen associated with skin oils and long‑term use, rather than a uniform varnish.
Painting and finishing: the bronze horseman
The bronze horseman follows a different finishing approach. Rather than painting metallic colour directly, the model is first coated in matte black. This acts as a shadow base, ensuring that any paint missed in recesses reads as depth rather than exposed plastic.
Metallic colour is then introduced using Antique Gold Rub ’n Buff, applied through a dry‑brushing process so that pigment catches only on raised surfaces. This technique works particularly well with the subtle grain of the SLS print, creating broken highlights that resemble cast metal rather than smooth paint.
Age and wear are introduced through thin brown and ochre washes, which are allowed to pool in crevices before being partially wiped back from high points. This layering process avoids a uniformly “dirty” appearance and instead produces variation consistent with long‑term exposure and handling.
A final light touching of Rub ‘n Buff to the tips and sharp edges of the horseman to add the glint to these ends.
Conclusion
The resulting prints are not intended to replace the artefacts they are based on. Instead, they offer a way to examine how historic objects can be translated into contemporary materials while still retaining evidence of use, age, and making.
A project to capture and reproduce an 18th-century coin from the collection of Michael Talbot, Associate Professor in the History of the Ottoman Empire and Modern Middle East at the University of Greenwich.
The original silver coin, just 28.7 mm in diameter, was transformed into a one-meter-wide replica carved from birch plywood.
Denomination: 1 Budju
Material: Silver
Width: 28.7mm
Weight: 10.37g
Mint: Algiers
Year: 1239 (1824)
Language: Arabic
The Arabic inscription reads:
“Sultan of the two lands and ruler of the two seas, the sultan Mahmud Khan, may his victory be glorious.”
Replicating the coin comprised of the following steps:
Capture the Image:
Take a high-resolution, evenly lit, perpendicular photograph of the coin.
The original coin would have been struck by hand, here it has resulted in a slightly misaligned strike that left excess material, or ‘flashing,’ on the lower edges.
Generate a Depth Map:
Convert the photograph into a grayscale depth map – an image where varying shades of black and white represent the coin’s relief (depth and height).
There are a number of techniques for estimating a depth map from a 2d image; traditional, rule-based / algorithmic techniques – and more recently AI neural networks trained on a huge datasets have proved very effective.
Some examples of these (open source) models are Marigold, Midas, DepthFM. There are online versions of these that can be used – but in this instance DepthAnything was used to process the image locally.
Create a 3D Model
Apply the depth map as a displacement map onto a flat digital model to create a 3D surface.
With the depthmap image as a PNG image, this can be applied as a displacement modifier in most 3D editing software then exported as an OBJ.
• Rhino – HEIGHTFIELD command, select the depthmap file. Increase the sample points to get a finer mesh
• 3d Studio – Select the geometry (cylinder) and apply DISPLACE modifier, load the depthmap image – use the strength parameter to control height of displacement.
• Blender – Add new mesh plane, subdivide the mesh in object mode many times, apply the displace modifier, create in that a new texture and open the depthmap image to apply it.
CNC Toolpath Generation
Import the .OBJ into CAM software to generate the toolpath – these are the digital instructions the CNC machine will follow to take the drill bit around the material fixed on the CNC bed.
CNC Machining:
Run the CNC operation to carve the 3D coin shape from a block of plywood according to the toolpath.
The CNC machine will perform a two-pass subtractive process: a roughing pass with a larger bit to remove the bulk of the material, followed by a finishing pass with a smaller bit for finer details.
The first pass took around 40 minutes, while the second pass took 6 hours to complete.
Finishing
Sand down any rough areas on the carved coin and apply Tung oil to finish the wood.
The CNC process can leave fine tool marks, and rough surfaces; sanding smooths these imperfections, revealing the wood’s true character. Tung oil, a penetrating oil finish applied over 3 days, soaks into the wood fibres, making them appear richer and darker. This brings out the grain and enhances the contrast and visual depth of the wood’s natural pattern.
After the Great Exhibition of 1851, the Crystal Palace structure was moved from Hyde Park to Penge Place in South London, now known as Crystal Palace. A railway station and an elaborate subway were also built for it.
The Palace itself burned down in 1936, and the railway station was demolished in 1961, leading to the subway’s gradual disrepair. Around 2010, the Friends of Crystal Palace, Bromley Council, and Historic England initiated a major restoration project. The subway reopened earlier this year.
The vaulted area is directly below Crystal Palace Parade and features 18 stone columns topped by terracotta-and-cream brickwork fanned vaults.
The busy road is about 1.5 meters above the vaults – buses can be seen on the road in this scan.
The same view in elevation
This section is made up of both scanner’s output – the BLK2Go for the closer / colour point clouds and the Hovermap for the longer range street captures
This is an undertaking to replicate an 18th century Ottoman gravestone which is currently held by the Royal Greenwich Heritage Trust.
Michael Talbot, Associate Professor in the History of the Ottoman Empire and Modern Middle East at the University of Greenwich, was informed by the Royal Greenwich Heritage Trust about an “Arabic tablet” which he identified as an Ottoman object and subsequently transcribed and translated the inscriptions on it.
The original tombstone itself is a late-eighteenth century artifact made of limestone or marble featuring Ottoman Turkish inscriptions in the sülüs calligraphic style. The gravestone’s origin is not clear, but it was possibly brought back from Constantinople as a memento by a British officer in the 19th century. The inscription features a poetic composition reflecting the youth and untimely death of its owner.
The tombstone itself measures 72 cm x 21.3 cm x 11 cm thick. The stone is broken at the bottom where it would have originally been set into the ground. It would also have been topped with a carved representation of the headgear associated with the deceased’s rank and profession, likely a turban, indicating a position in the religious-scholarly class.
Replicating the tombstone comprised of the following steps:
Scanning: photograph all sides / angles of the object and utilising photogrammetry to generate an accurate 3d digital model of it.
3D Printing: produce actual-size moulds of the tombstone from the scan model.
Casting: pour Jesmonite (similar to plaster) into the moulds and allow to set
Scanning
Photogrammetry is a process of 3d scanning whereby many photographs of an object are used to create an accurate digital model. Common points in the overlapping photos are identified in order to align them and create a point cloud – a 3D representation of the model as dots in 3d space extracted from these aligned images. This is further refined into a mesh model to form a network of triangles which are lastly “wrapped” with the texture that has been derived from the photographs to provide colour and detail to the 3D surfaces.
To get the best results for use with the photogrammetry software there should be many photos which are sharp, evenly lit – with as little shadow as possible – and capture all sides and angles. Ideally the object would be photographed in a photo-studio with controlled lights and blank backdrop etc, but since this was not possible in this case some soft lighting was brought to site to offset directional shadows from windows.
The model was photographed with a Panasonic Lumix FZ82 (on a tripod) which is a mid-range bridge camera. Manual settings / RAW format, approximately 500 photographs – then post processed in Adobe Lightroom to eliminate any blurred shots and batch edited to further reduce shadows and bump up highlights.
The photogrammetry software used for this exercise is called Reality Capture. Since the tombstone was too heavy to stand up on its end it had to be laid flat, horizontally for one set of photos and then flipped onto its front to capture the other side. Ideally Reality Capture would have automatically detected all the photos as a single object but in this case it generated two separate components: a top and a bottom. To fix this one halve had to be flipped and manually aligned in the software in order to produce a single complete model.
With the model successfully generated it can be exported to a variety of formats for a variety of purposes. For viewing / zooming / spinning the model online it has been exported to Sketchfab this model includes the texture for added realism.
For purposes of 3d printing the model is exported to a common OBJ file format. The texture wrapping step is not important to 3D printing since these printers do not reproduce the model’s colour, so the version used there is effectively monochrome.
3D Printing
Commonly 3D prints are created using PLA or photopolymer resin etc. While these materials recreate accurate models, they can feel light and “plasticky”. For the tombstone it was important to recreate as much as possible the tactility of a stone / marble material and have a weightiness that approaches something more authentic to the original in that respect.
For these reasons instead of printing the model directly, moulds of the model – effectively an inverted version of the 3d scan – were produced, into which a plaster-like material was poured and left to set.
Material used to print the moulds: TPU 95A flexible filament. This means the resulting structure is supple and bendable – strong enough to hold the pour but can be peeled away from the cast once it has set.
3D Printer used: Bambu Lab P1P. The maximum printable area of this printer is much smaller than the size of the tombstone itself meaning the mould needed to be printed as 4 sections and then reassembled for casting. Each of the sections took 18 hours to print. (shorter test section illustrated above with red mould).
Since a full dense model would use a lot of casting material – and also create a very heavy model – only an outside skin of about 10-15 cm of the model was needed to be cast. To achieve this an additional 3d printed core of the tombstone was placed inside the mould in order to cast around it. The final model is lighter and more economic with casting material and retains the proper look and feel – but with a hidden, enclosed, non-dense core.
Casting
The material used to cast the tombstone is Jesmonite – this is a water-based composite material that combines natural materials such as gypsum and resin with various other components, including water-based polymers. It is known for its flexibility, durability, and environmental friendliness compared to traditional resin-based materials. It is also less likely to chip or crack like regular plaster and can be mixed with a pigment to colour the material.
Images below show Jesmonite being mixed with pigments and some experimental, test colours.
The 4 separate 3d printed moulds were taped together to form a continuous mould to be cast into – so the moulds were printed as 4 sections with the casting as a single object which required no reassembly.
The photo above left shows the 3d printed, light core inside the mould while on the right the Jesmonite is poured in to fill the mould with the core embedded inside.
The cast itself sets fairly quickly in about 20 minutes and is then ready to be removed from the mould. The moulds themselves can be re-used to produce more replicas (though the core would require re-printing).
The weight of the replica is initially about a third of the weight of the original, though as the moisture evaporates over a few weeks it becomes somewhat lighter, though still a substantial weight of about 15kg.
A scan of 23 stone heads stored at the Old Royal Navel college in Greenwich.
Carved around 300 years ago by sculptor Robert Jones, around 50 of these heads were commissioned as decorative features for the courtyard of the Painted Hall – originally intended to be built in stone. A financial decision to swap to brick instead meant the stone heads were never used.
The best preserved (and repaired) heads are now on display in the ORNC visitor centre – with two more set into a fireplace in the Queen Mary undercroft. The rest that survive reside in the vaulted undercroft of Queen Anne and make up this scan survey.
Made from Ketton limestone the heads represent mythical characters of the sea (with the addition of British Lion heads). They are stacked up, chipped – some of them quite badly – and staring into the darkness; a haunting sight – and great material for scanning and digital manipulations.
The image below is a laser scan of the vaulted undercroft where the heads are situated – a remnant of the old Tudor Palace of Placentia.
A visit to a vast disused factory in Paris before it is demolished. The old Pouchard Tubes pipe factory was a historic industrial site in Pantin, a suburb of Paris. It was founded in 1927 by the Pouchard family, who specialized in cold deformation and surface treatment of steel tubes for various applications, such as automobiles, boilers, and railways.
Les Grandes-Serres de Pantin: Truss / Gantries
The factory occupies a large area between the railway tracks and the canal de l’Ourcq, and has distinctive glass-roofed halls that gave it the name of “Les Grandes-Serres” (The Great Greenhouses)
Les Grandes-Serres de Pantin: Main Hall
The factory itself closed in 2017 and the site acquired by a real estate developer with plans to transform it into a mixed-use complex with offices, services, shops, restaurants, and a hotel.
Les Grandes-Serres de Pantin: Rotating Gantry Rig
Prior to commencement of the redevelopment the premises were made available as a huge artist space for exhibitions and installations; this scan survey was carried out in cooperation with “Boite à tubes #1, Grandes-Serres de Pantin, 2021”, an electroacoustic work on the industrial sound memory of the site.
The installation was created by Nadine Schütz, a Swiss artist who works in sound and space. She created the work in collaboration with the artist collective Echora.
Les Grandes-Serres de Pantin: Houdini Volume Light EffectsLes Grandes-Serres de Pantin: Houdini Point Size Experiments
The sound installation itself is located in a small control cabin in the Grande Halle of the Grandes-Serres which has been perforated with a variety of metal tubes of different sizes. Each tube plays a different sound, creating a cacophony of industrial noises which act as powerful reminder of the industrial past of the Grandes-Serres site.
Les Grandes-Serres de Pantin: Adjacent Row of Lime Trees
While the many lime trees that surround the site cannot be relocated, the intention is for their sculpture-grade wood to be repurposed within the new development.
Les Grandes-Serres de Pantin: Cyclone 360 Orthographic Projection
The Brunel Museum tunnel, also known as the Thames Tunnel, is a museum in Rotherhithe, SE London, consisting of the engine room and the shaft down to the tunnel.
Built by Marc Brunel and his son Isambard Kingdom Brunel, The Thames Tunnel was the world’s first underwater tunnel for pedestrians. When the tunnel opened in 1843, it was hailed as the 8th Wonder of The World.
By employing a pioneering tunnelling shield and other novel techniques, the Brunels encountered waterlogged ground, fires, and the ever-present threat of the River Thames above. Their success paved the way for future underwater tunnels and underground transportation systems.
The shaft is now sealed over, as the tunnel is now used by Overground trains – though the trains travelling between Rotherhithe and Wapping stations can still be felt and heard.
These two images illustrate the approximate location of the the train tunnels below the shaft seal.
Animation of Ortho camera descending through the site
