Real-time filmmaking is coming… most won’t survive it
Artificial intelligence is here, and it will fundamentally alter filmmaking. But, says The Moon Unit, AI's democratisation of the industry is a fallacy, because as the jigsaw of real-time filmmaking edges closer to completion, it will be the select few with access to the knowledge, rather than the masses with access to the tech, who will come out on top.
)
Live coder Switch Angel generates high-energy EDM effortlessly in real time [below]. No UI, no fancy VST plugins or built software, just pure, juicy lines of code being written in the moment, forging beautiful progressive trance melodies. It’s here, it’s happening, and it’s utterly mesmerising.
But what of visuals?
Despite all those compelling Instagram claims, AI isn’t about to turn everyone into a great filmmaker.
Real-time AI filmmaking is no longer hypothetical, it’s imminent. But, as the tools accelerate, the myth of effortless creativity collapses. In a world where anyone can generate images, only those with deep, hard-earned expertise will know what’s worth making.
Above: Content creator Switch Angel shows how the combination of talent and real-time creation can make beautiful music.
The expert dreamers
Let’s kill the fantasy early. Despite all those compelling Instagram claims, AI isn’t about to turn everyone into a great filmmaker. It’s about to reveal - very quickly and very publicly - those who never were.
Real-time AI filmmaking is coming fast. Not as a distant, speculative future, but as an emerging reality. Within a few short years, directors will be able to generate scenes on demand, shape performances live and iterate visuals, sound and tone in real time. The production pipeline as we know it - shoot, edit, grade, VFX, sound - will collapse into a single, continuous act.
Real-time AI filmmaking is coming fast. Not as a distant, speculative future, but as an emerging reality.
This may sound like democratisation but, actually, there’s a greater purpose; it’s a filter. What real-time filmmaking actually means is that, if you strip away the hype, you’re left with something way more radical than 'faster production'. Real-time filmmaking means this:
A director inputs a prompt - typed or spoken - and a scene appears. Not a rough sketch, a fully realised moment: lighting, performance, camera movement, atmosphere. Beyond simple generation, this is deep interaction.
The director refines the scene live:
- Pushing from a 35mm to a 50mm lens with a gentle pull out.
- Pulling back performance - less overt, more internal.
- Shifting colour temperature to something colder, more distant.
- Tightening the pacing, trimming emotional excess.
The system updates instantly, no renders, no delays, no safety net, just precise, informed decisions based on deep knowledge.
Above: Real-time filmmaking will become a filter for those who have the knowledge, and those who don't.
The building blocks are here
All the pieces already exist, they’re just not yet fully integrated.
- Real-time engines already deliver near-final imagery live.
Generative AI models are producing moving image with increasing coherence. - Prompting is evolving into a new directorial language - part screenplay, part shot list, part emotional brief.
- Synthetic performance systems are beginning to allow control over expression, tone, even nuance.
When these systems converge, filmmaking stops being a process and becomes a live orchestration layer.
Campaigns become dynamic systems, not fixed assets. Variants are generated at will.
You get:
- Live cinema: scenes shaped in the moment, not assembled later.
- Infinite iteration: ten versions of the same idea in minutes.
- Radical compression: smaller teams, faster output, higher volume.
- Adaptive storytelling: narratives that can flex, respond, evolve.
For advertising, the implications are obvious. Campaigns become dynamic systems, not fixed assets. Variants are generated at will. What that really means in practice is a shift from campaign delivery to campaign evolution. A single idea no longer resolves into one hero film and a handful of cutdowns, it becomes a living engine, endlessly adaptable.
Different audiences receive different tones, performances, even story outcomes, all derived from the same core concept. Media buying and creative begin to merge. The question is no longer 'what did we make?' but 'how intelligently can it respond?'.
Above: When the pieces finally all come together, what you get, says The Moon Unit, is filmmaking as a "live orchestration layer".
This may all sound good so far, but there’s a problem; when everything becomes possible, most of it becomes worthless slop.
The constraints (for now)
Before we get carried away, there are still key barriers:
- Maintaining character consistency across long sequences.
- Sustaining narrative logic over time.
- Achieving precise, repeatable control.
- Managing the cost of real-time, high-fidelity generation.
These will be solved and, when they are, the technical advantage disappears. All that’s left will be taste.
For over a century, filmmaking has been about capturing reality, even when stylised, and now we’re generating it. And what a world it is; no camera required, no location constraints, no physical limits.
If you can make anything, instantly, the only question that matters is: Why this?
The prompt becomes the camera. The system becomes the set. The director becomes the end-to-end decision-maker. But this is where things get uncomfortable, because if you can make anything, instantly, the only question that matters is: Why this?
Above: If you can make anything, the main question becomes, what do you make?
Enter the expert dreamers
This is the part which the 'anyone can do it' narrative conveniently ignores. The future doesn’t belong to people who can prompt, it belongs to people who understand what they’re prompting for.
In a real-time system, the machine doesn’t create taste, it exposes it.
The 'expert dreamers' are defined not by tools, but by depth:
- A working instinct for cinematography; not just lenses, but intention
- A granular understanding of performance; when to push, when to restrain
- A lived knowledge of film, advertising and music video history; as language, not reference
- Fluency in fashion, design and architecture; because image is meaning
- Sensitivity to sound; because emotion is often heard before it’s seen
- Mastery of structure; pacing, arcs, tension, release…
Because in a real-time system, the machine doesn’t create taste, it exposes it. Ask for something generic, you get something generic… instantly. Ask for something derivative, you get something forgettable faster than ever before. There’s no post production to rescue you. No edit suite in which to 'find the film'.
You are the film. This is now your performance.
Above: AI isn't levelling the playing field, it's tilting it sharply.
A timeline (and a warning)
Based on current trajectories, the near future might look like this:
- 1–2 years: high-quality, near real-time scene generation becomes usable.
- 3–5 years: full sequences, stable characters, live iteration tools.
- 5–10 years: true real-time filmmaking environments driven by experts with deep domain knowledge.
At that point, the question shifts from 'can we do this?' to 'who actually can?'. Access will be universal, but ability won’t be.
Above: At some point, the question around AI won't be, 'what shall we do?', but 'who can do it?'.
The conclusion no one wants to hear
AI isn’t levelling the playing field, it’s tilting it sharply, because when execution becomes instant, the only competitive edge left is judgement; the accumulation of everything you’ve seen, studied, absorbed and understood over years.
In other words: craft.
So, yes, anyone can now generate a shot, but making something that lands, something that moves an audience, sells an idea, earns attention in a world drowning in content and increasing amounts of slop? That still requires something rarer. Something slower to build, something harder to fake - something that’s impossible to automate.
AI isn’t levelling the playing field, it’s tilting it sharply, because when execution becomes instant, the only competitive edge left is judgement.
So, here’s the real question for this industry: Are we training a generation to use tools or to develop taste? Because one of those scales, the other survives.
Need inspiration? Imagine the skillset needed to develop the above, or this…
If you’re a director, a creative, an agency leader then now is the moment to choose. Go deeper. Study harder. Broaden your references. Obsess over performance, structure, image, sound, culture.
Build the kind of creative intelligence that machines can amplify but never replace, because when real-time filmmaking arrives - and it will - the gap won’t be between humans and AI, it will be between the expert dreamers who have something to say and those who finally realise they don’t.
