The work I can show you, in the order it happened.
A good deal of it never appears anywhere — the consulting, the products that shipped under somebody else's name, anything still under agreement. These are the highlights I am free to describe.
Founder and CEO for two decades. In 2001 the software that made films look like films was built for facilities and priced for them. The independent filmmakers with the most ideas had the least money. I set out to help people find their voice and enjoy making films. That turned quickly into something bigger: professional editing, compositing and VFX within reach of anyone who needed them, and the teaching to make the most of it. A multi-award-winning company, both for our products and our customer service.
The simple visual-effects product that started it all, built on the effects people loved from Star Wars and The Matrix — and the community of young filmmakers around the world who found a home around it.
Many of them went on to work on the real Star Wars films.
Effects plugins for Apple's iMovie, built at Apple's invitation to widen what their new editor could do.
Shipped before FXhome had its name.
AlamDV's successor, and where the effects stopped being approximations. Lightsabers, muzzle flashes and the rest, rebuilt for fidelity.
It also opened up: the community could build their own effects and add them to the product.
The keying product, on Mac and PC, carrying our latest video technology — award-winning chroma key that held up on the cheap cameras people actually owned.
That keying went on to underpin VisionLab and PhotoKey.
EffectsLab and CompositeLab in one product, with a great deal more besides, and a teaching platform built alongside it. That combination is what turned an audience into a grassroots movement.
Award-winning, as a visual-effects tool to run beside whatever editor you already had.
The world's leading green-screen photography and event software: shoot against green, composite and deliver on the spot.
Used from theme parks to Formula One and the Wimbledon Tennis Championships, and white-labelled into kiosks and booths around the world. Award-winning, as the best green-screening tool on the market.
Professional post-production, priced so a sixteen-year-old could learn on it.
Everything the earlier products had taught us, rebuilt as one thing: editing and visual effects in a single 3D space, with compositing that went further than Adobe’s own at the time. 3D particle systems and models, keyframed and behaviour-based animation, colour grading, and an editor quick enough to actually cut on.
The entry version went out on pay-what-you-want, with the option to send your payment to charity rather than to us. Our customers being who they are, they took it — extensively. Six million people called HitFilm home, and the community they built around it was extraordinary.
HitFilm's effects technology, unbundled from the application and shipped into everybody else's editors.
Tens of millions of people — some through direct sales, more through being bundled with some of the biggest video products in the world.
The first non-destructive RAW image compositor: HitFilm's thinking applied to high-resolution stills, with layers that never touched the original.
Compositing beyond what Photoshop allowed.
Virtual production that fits in your pocket.
CamTrackAR recorded the camera’s path in 3D while you filmed on an iPhone or iPad — handheld, because the track came off the device rather than a rig — so the tracking data arrived in post alongside the footage, with green-screen capture and previsualisation you could see on location rather than imagine. Apple featured it in the 2021 iPad Pro launch keynote.
Then it went the other way. Something one person could shoot handheld was picked up in Hollywood, for stunt coordination and pre-production on Disney and Marvel films.
Through the acquisition and five years inside it, latterly as Global Chief Innovation Officer. Generative AI was the work: the research, the demonstrations, the prototypes, and the working products that came out of them. The public part is below.
Colour grading and VFX plugins for the major editors, included in the subscription rather than sold alongside it.
Every speech model worth testing, trained on our own material and put in an arena to compete — judged on performance rather than on how clean it sounded.
Naturalness fails in specifics: breath, hesitation, where the emphasis lands. A creative team could hear one model against another and choose per voice and per language, instead of taking one vendor's word for all of it.
Stills, and the part that decides whether they are usable: not only generating a picture but being able to edit it afterwards.
Style, lore and prompt adherence were all measurable in the arena. What separated the models was what happened after the first result — an image you cannot change is a result rather than a shot.
Video, where the hard part is never the first shot: keeping the same character, the same location and the same look from one to the next.
Consistency is what separates a demonstration from a sequence somebody can cut, and no benchmark reports it. The models were compared on it directly, frame to frame.
A cinematic model of their own: training, lore and prompt engineering layered across the state of the art, so the flexibility and the adherence are there and the picture still looks like nobody else's.
Flexibility, prompt adherence and quality are the entry price, and every model is converging on them. A visual language of its own is the part that cannot be copied out of a benchmark — and it is the part a creator can actually direct towards something.
The first problem I ever worked on, with much better tools.
Framing a painting is a decision you make once. The moulding is cut, the work is mounted, and changing your mind means starting again on an object that is often irreplaceable. For most of the trade’s history the way to make that decision was a corner sample held against the canvas, and a good eye.
I first worked on this at eighteen, with my father, on software that let a client see a painting in a frame before committing to it. The Times covered what we were doing in 1996. Nearly thirty years later I came back to the family business and built FrameEdit on the same idea: high-fidelity rendering, a large frame library, and AI that can identify the artwork it is looking at.
FrameEdit was used in the process of framing René Magritte’s L’empire des lumières, which sold in 2024 for $121.16 million.
Advanced green-screen keying in After Effects, based on the technique my friend Niko Pueringer of Corridor Digital developed. I built the port: drop it on a layer, no setup, no cloud, no account, and the same result on Mac and PC.
Needing no technical skill was the brief, not a feature of it. The technique already existed and was remarkable; what did not exist was a version an editor could apply to a layer and get on with the shot.
Advising the team building Everslate.io: infrastructure, tech stack, and getting out of AI the results they need rather than the ones it offers.
A platform for documentary producers and commissioners.
Integration, platform and launch work across two decades of building creative software. Relationships, not a client list.
Apple chose us to showcase AR filmmaking at the 2021 iPad Pro launch.
START A CONVERSATION
Tell me what you are weighing up and I will tell you what I would do — including when the answer is that you do not need me. Three or four sentences is plenty.