I’m currently available for 3D generalist work in the UK, ideally either remote, or as a hybrid to London arrangement. I do a lot of layout and environments, asset creation, lighting and leading small teams.
A Gentleman In Moscow is the last show I worked on during my latest stint at Rumble VFX. The reel above contains breakdowns of many of the shots I worked on.
Creating a City
I was initially tasked with creating a reusable Moscow skyline asset. This was inspired by photographs of the 1920s and 1930s. Many of the buildings used were from off the shelf kits. These had textures, but weren’t yet set up for Redshift 3D, nor indeed Houdini ready. The process became much easier after a few models. For example I had a script set up to convert shaders to Redshift ones with the correct names and textures attached. All needed balancing to look approximately the same when brought in to one primary scene.
Photography of the time shows the typical Russian formal gardens in front of the Bolshoi and Metropol. I began the layout with these gardens, working out from there. Much of them are point instanced models of bushes, flowers and grasses with hero shrubs and trees placed by hand. This was in part because in some shots they got in the way and in others they needed adding to so as to give height to the scene.
Paved areas were defined with curves. Stoned edging and cambered road surfaces were added in procedurally along and away from the curves respectively. There was even a tram system for creating recessed metal rails in the road and slightly different cobbles on the ground. A separate one defined the power lines. In one shot I went to the extent of animating the over head lines and the pantograph on the tram, something that is maybe only visible on very large TVs!
Optimisation
A technical challenge with having so many models and lights in shot, was that of memory limits, especially VRAM. To remain quick at rendering Redshift must use less than the available GPU memory. Plants were read off disk at render and most of the buildings were converted to rs proxy files. They load in quickly and could still have their shaders altered globally for one specific pass, the snow.
The snow pass was in addition to the usual beauty with a snow material overriding all the usual ones. It was blended with the beauty using Normal, AO, noise and other custom AOVs in Nuke. This method allowed us to be flexible about the location of snow. As with other projects where I’m setting up a primary setup for a team I had a Nuke script that could show compositors roughly the end result I was aiming at.
Snow on the ground was actually one large grid with its own gentle undulations, pushed up near to buildings, street furniture and so on. It was then roughened in these locations and poly reduced, keeping the detail in these bumpy areas. To add further detail when seen at ground level, the snow grid was worked in to with displacement maps, particularly footprints.
Smashing work
Other tasks on A Gentleman In Moscow included animating a picture frame, shot by the count to alarm the hotel manager. In the plate, the practical gun went off then half a second later the picture swung down! For impact, timing and believability we decided to replace the whole thing post-firing. I made a simple model of the frame, complete with photo, mount board, glass and frame, then I set up a shattering system.
Stills on set gave me a great reference for which bits of glass remained in the frame and where the bullet hole should be. I drew out wobbly lines radiating from the hole and set up an RBD sim in Houdini. I had trouble getting the finesse I wanted from the SOP level shattering system so only used that to break the mesh in the first place, passing that and the constraints to a custom DOPnet. In there I gave the sim a speed limit, some drag and transferred velocity from the picture frame over the first frame or two for pieces within a certain range. This was to assist with the appearance of glass being pushed by the frame as well as the bullet. Anything that shot backwards for any reason was rapidly decreased in size so as to vanish.
The last few years have seen a lot of changes to my reel, though a couple of projects are missing, either through incongruity or tight Netflix IP restrictions, so I’ve snuck in some pre-pandemic stuff. I’m working on a private reel including shots from The Crown, so if you’d like to see that, get in touch via aj@ajcgi.co.uk.
Like many in the UK VFX scene I still work either remotely or hybrid. My commute in to London is a 4 hour round trip so please take that in to account if you get in touch after seeing this reel.
Primarily I use Houdini. There are some who would claim that’s not the weapon of choice for generalist work, but I would contest that. Lots of my work is documentary and drama content where I’ve been hired to do environments and set extensions. This means I do asset builds, scene layout, lighting, and just about everything including tracking and sometimes the compositing too. Houdini works fine for practically everything I throw at it and is flexible enough for me to set up entire scenes as a base for others and have them understand what they’re looking at.
Naturally there’s other work absent from this 3D VFX Showreel, so take a look at the projects page for more.
In 2023 I was brought in to Rumble VFX as a lead 3D artist on Raw TV’s documentary series, Lost Cities Revealed.
Each episode covers a different location where up to date archaeological methods have revealed settlements to be larger and of greater importance than previously thought.
The VFX task was to create a big moment at the end of each of the six episodes, the current landscape altering to show how things may have looked in the heyday of each civilisation.
The Scottish episode focussed on Pictic Celts, a civilisation a few of us had experience recreating together for Britannia many moons ago. In a sense that made it easiest to make it look convincingly real. It probably helped I holidayed in the Highlands immediately prior to working on the series too! Little circular huts, wooden fences, fires, smoke, a suggestion of farming all helped to set the scene and give atmosphere.
Oman’s episode focusses on the first ever trade post at an oasis, an area of water-fed rich soil in an otherwise desert landscape. For this one I had to create buildings based on a few drawings and the excellent feedback from the director. The actual archaeologists on the ground had moved on to another project and didn’t have ready access to comms. Having made a generic square hut with a procedural redshift material for the mud walls, I laid out a city based on a map of the area, featuring irrigation and date palms.
Each episode came with a LiDar scan of the site, produced by industry specialists, Visual Skies. The Oman one really helped as the 2nd shot transpired to not be wide enough to fit in the oasis itself during the reveal. I replaced it with an entirely CG shot, recreating the camera move, only higher and wider.
As with all documentary work I’ve undertaken there’s a strong element of artistic license, be that in the layout of the cities, the overall look of buildings compared to each other or colours used by civilisations. In short, research is key when working on scientific or historical stuff, but where evidence is lacking, artistic expression helps.
The video above also contains shots from Sudan and Mexico.
Lost Cities Revealed is on Nat Geo and Disney Plus. I’ve done a lot of documentary work. To see more, take a look through the Projects page!
I’ve been fortunate enough to keep working with Rumble VFX for a while now, recently completing a couple dozen shots on Season 6 of The Crown, the final instalment of Netflix’s favourite. Our work, alongside Framestore’s, has been nominated for a BAFTA Craft Award!
I began this stint on The Crown making many heraldic banners for St. George’s Chapel, Windsor, filmed on location in York Minster. The real chapel has many heraldic items which you can read about here. Of all the heraldry in the real chapel, the banners are visually the most impactful.
From a technical perspective, each banner is connected to a point on a line. Each point has variables to define the angle the banner droops at, plus some colour variance. A vellum cloth solution is applied to all flags at once in Houdini, with the fringing being extruded from the edge of the resulting mesh.
As I knew the order of flags, and indeed gaps between, were subject to change I renamed all the textures and shaders such that they’d pull in flag_1 textures for the first flag, flag_2 for the 2nd etc. This afforded the flexibility to merely rename the textures to change the layout.
Diana
The investigation into Diana’s death took a very long time, wrapping up practically a decade afterwards. As part of a montage surrounding the crash investigation, we see a computer simulation of the car passengers during their moment of impact. To give the impression of a ragdoll-like movement inherent in these things I used a ragdoll solver within Houdini on characters with very basic wireframe appearance, sat within a model of the Mercedes.
Many different simulation systems in 3D share a common trait – one tweak to a shot can change the whole thing, with characters falling the wrong way, leaping out their chair through the sunroof or slapping each other in the face. There were a couple of occasions where the motion appeared too violent, even given the wireframe appearance. With each revision we lost more and more visual detail, dropping frames to give it more of a caching computer simulation look and less of an emotional re-enactment.
Half a Dimension Lost
An odd concept I first read about a decade ago is 2.5D, that is to say a 3D look made in 2D. As I was tracking a few shots for a sequence of princes Harry and William having fun in Botswana I decided to drop in a temporary background in Nuke as a test. It made sense to use the digital matte painting provided and after a few tweaks and advice from others I realised that projecting that on to geometry in Nuke wasn’t so tricky. All my Botswana shots were Nuke scanline composites, not 3D renders. To me it pays to know a few little tricks involving multiple software. I’m hired to produce the end result not a convoluted Houdini scene nobody except wizards can understand.
My work has seen me do many things in many ways – see other projects here.
A while back I wrote a blog post on my photography site about AI image generation. I was wondering how long it would be until AI could create a convincing photograph. Fast forward to today and Photoshop has AI tools built in with Adobe creating AI video tools for After Effects and Premiere too, all with convincing results. There are many, many others doing the same.
Strikes
It’s understandable that writers and actors have been calling for strong protection against their own writing, voices and even appearance being used without fair recompense. At the time of writing, American unions have been striking for over 110 days for many reasons, but featuring in them is a desire for fairer AI protection. This lengthy period has led some to endure loss of income, jobs and even homes across California. It’s had far reaching consequences worldwide in fact, with plenty of UK projects being suspended or pulled thanks to the links to the US.
I’ve been thinking about this for a while. It’s easy to shout out and tell folk to get back to work. Imagine though, not having that desired protection or framework in place. In that scenario, the current 10 figure cost to California alone could potentially be far higher.
Actors and writers could lose out to an ever decreasing pool of AI-assisted talent. Voice actors might find their voice effectively bought with unfair terms in perpetuity. Their income would be gone. Actors could be placed in any video without permission, causing a Wild West of IP lawsuits.
Productions themselves could get weaker, with script after script of middle of the road shows, AIs borrowing sentence structure and characterisations from years of work. Visuals could homogenise, with actors turning in to uncanny-valley dwelling automata.
Even if this dystopia I paint here is temporary, the loss of earnings could be huge and quality of work underwhelming compared to the original content made by real humans daily.
A brighter future
The TV and film industry will likely shrink in light of the strikes and common usage of AI. It has appeal for small graphic design projects and concept work. I do also believe though that we are an incredibly diverse, creative species, capable of creating extraordinary worlds and telling beautiful stories. We are also able to diversify in what is a fast-paced industry.
I personally believe many VFX jobs are safe. Directors, producers and editors all like to put in their opinions on how a particular VFX shot can be improved. The first version will go in a creative direction they may or may not like but it will need iterating on usually. Many of the requests made are open to interpretation and fairly abstract. If you want great results out of AI, it pays to be specific.
Many software developers are integrating aspects of machine learning into their VFX software, making awesome tools that make our jobs easier. AIs are not likely to create entire shots themselves that fulfil a client brief first time, if at all. Let’s hope the strikes end soon so everyone can use these awesome tools to keep making original content that folk want to watch and those who need protecting feel safe in their chosen work.
CGI – Computer Generated Imagery, as a term has always been an enigma. Personally I see it as being 3D graphics, but Wikipedia confusingly refers to it as including some 2D also.
VFX – Visual Effects, to many is synonymous with green screen. I’ve long suspected this belief comes from Behind The Scenes documentaries, vacuous DVD fillers that are so popular I do wonder if many people prefer so-called BTS content to actual movies and TV shows.
The VFX industry is so vast in scope that keying green screen (or any colour really) is one of dozens of things you could be asked to do within a week as a compositor. In itself, compositing is one of dozens of jobs in the business. However, very little compositing work is what many regard as CGI.
There! Right… there! You see it? That’s the grey area. If a viewer of a cinematic spectacle sees a hint of VFX they may well jump to thinking of it as CGI. Not all VFX contains CGI. Am I splitting hairs because I do 3D CGI and see CGI as 3D only? Yes. Yes I am.
When people say that show X has no CGI in, they might be right, or at least think they are. Lots of shows I work on have literally hundreds of VFX shots but only a dozen or so contain 3D graphic elements. The amount of VFX work in TV is astonishing. If you don’t notice it, it’s succeeded in being excellent.
To me, traditional 20th Century Hollywood was about in-camera practical effects and hand-painted backdrops. These days those are often, but not always, referred to as SFX – Special Effects. That historically has been used for so many things that now even VFX is lumped in with SFX in the media, to the extent awards are given out in the category of Special Visual Effects. Add in AI imagery and now nobody outside, or indeed inside, the VFX business has a clue what to do with all the acronyms.
The recent Barbenheimer furore made me think. Both Barbie and Oppenheimer contain a lot of VFX work – tonnes of it – but the directors’ favour of traditional methods was set upon by the media as a good thing, getting us all away from that pesky CGI. The CGI was never the issue. It was scriptwriting, acting, the terrible art direction, but most of all, a complete and utter misunderstanding of the whole VFX business and those who work within it.
To me, films and TV would improve a lot if the focus returned to making gripping stories with well-developed characters. Get that right, then speak to a VFX studio about what might work best as practical or VFX work. Read around the subject, talk to us VFX folk directly about what we’re doing, (and credit us if you would be so kind) but leave those misleading BTS docs alone. They aren’t made by those who made the effects.
To see the things I’ve worked on over the years and judge my qualifications for judginess, see Recent Work.
For me, there was a lot of set extension work to do including a recreation of the famous 90s neon signs at Piccadilly Circus. Naturally, being The Crown, there were also ground level, rooftop and aerial shots of Windsor Castle. I created a dilapidated look for Villa Windsor (Mohamed Al Fayed’s gift to The Queen) and was even called upon to replace an errant non-royal yacht. (How dare the late Steve Jobs leave it in shot!)
Work on the show took lots of hours of research and meticulous attention to detail, creating 3D in Houdini, projections in Photoshop, painting up in Substance Painter, rendering in Redshift.
A challenge for me is that I really enjoy the show and wanted to work on it for years, so ended up treating each shot as if it was my last. At one point I was dreaming of various tones of wall in the shots of Windsor Castle. What helped immensely and stopped me painting myself in to a corner was the exceptional production team whose feedback and documentation of the shoots was on point.
The yacht was a peculiar beast. Yachts are often very smooth, white, shiny, looking like fresh CGI frankly. With this being an HDR project I had to make sure we had details that were matching the plate, even out of the range of the SDR monitors most of us work with daily. When doing my rough comps I knocked the exposure of everything down to check it matched, then brought it up again.
One aspect of this project that really helped is the mountain of photos out there on the web. I’m really grateful for those of you who visited Piccadilly in the 90s with a camera or indeed the millions who’ve documented Windsor and the show’s Windsor, Burghley House over the years!
Set work is just one string to my bow – see other projects here.
For the last nine months I’ve been working for Fluid Pictures on graphics for Raw TV’s Gold Rush, a documentary series for Discovery Channel.
There seem to be two responses to the above. One is to ask what the show entails, the other is to ask why on Earth it needs visual effects. I tend to think of the Gold Rush work to be 3D graphics rather than VFX as it clearly isn’t aiming for a photoreal aesthetic. Unfortunately it’s this that keeps the show off my reel as the two styles clash. Luckily for me, Fluid have their own reel of graphics we’ve made!
I worked on many of these shots, bar the ones at 00:11, 00:17, 00:20, 00:26, 00:31, 00:42, 00:50, 00:53, 00:59, 1:03
I’ve personally been involved with season 10-12 and even a few shots on season 9. Each season is around 24 episodes long, with earlier episodes being broadcast as the later ones are edited. This time constraint, with a delivery every week, means graphics can’t be too fancy or they would continually fall very short of the client’s ideal.
That all being said, we used Houdini as the 3D software of choice, usually used as a straight VFX tool. The node-based methodology, excellent terrain tools and a fairly logical workflow worked in our favour. It was the first time where locations could easily be referenced directly from GIS data. We made tools for drawing out rivers, the cuts in the ground, tree distributions and so on, such that we could concentrate on the shot content. Much of the time the graphics explain how things work or the challenge of moving material from one location to another, but also mechanical things break, especially those with moving parts, so many of the graphics are there to show the problem and how it’s solved.
Oddly there are still technical challenges on this show, with Houdini FX actually being necessary for smoke. Also, although we had simpler Houdini Core solutions for dirt, water and conveyors, we needed to show these things going wrong dynamically or illustrate a point close up.
The show has improved my Redshift 3D skills and made me learn about HDAs, rigging excavators, POP fluids, and machining engine parts using VDBs. That being said, three series is enough for me now. I’m moving on to pastures new with actual VFX work on actual plates.
Gold Rush, at times in its life the most watched show in America on a Friday night, is available on the Discovery network of channels and Discovery+ in the UK.
After a few more years of pootling about working in London and 2 years of Houdini work, it’s about time I updated my VFX reel! The previous one missed many projects out, so it’s fitting that this one is practically a 2019 one, especially considering the months spent during lockdown working on a personal project or two.
This particular reel contains some of the more effective and impactful shots I worked on for The Planets and the second series of Britannia, both at Lola Post Production in London’s West End. The Planets mainly involved decorative spheres in space, with a strong style that leans heavily on NASA’s archives with inky blacks and no stars. See my preview post from a while back for details!
As an aside here, as someone who has switched 3D software to Houdini, if you’re considering learning Houdini, don’t be daunted! Start with the simpler stuff. At Lola, I was given a Houdini Core license which gets used in studios to do the day to day 3D tasks, working on shots, bringing in assets others have made, plus creating shaders, doing layout work etc. If you can handle that, the FX stuff becomes a lot easier to get your head around because you are already thinking Houdini. Looking back, especially now I have my own Houdini license at home, if I had dived into the FX end of things first it would have put me off the software. I can now easily work around problems with wrangles, writing my own nodes in VEX, I understand the logic with which the transforms are put together, the reason why global transforms are hardly ever differentiated from local and so on. If I had to learn that AND how to make a custom destruction sequence I’d be a full time bowl carver by now.
Enjoy the reel! And the drum and bass. Apologies, I needed something with pace and no lace. Feel free to use that last sentence in a conversation today.