Hooroo Jackson

The Infinite Protocol

The Grammar of Machine Cinema

How agentic filmmaking enables cinema’s modernist moment

I.

Cinema used to be organized around scarcity of the image. Machine cinema is organized around abundance of the image. This fundamental inversion changes every downstream rule. Instead of an image obeying linear logic, from shot to shot to construct a narrative, it may operate in service to a conceptual whole. The image no longer dictates logic, the idea dictates the concept and the image reflects the preordained question.

I call this the Infinite Protocol: when any shot, any performance, any aesthetic choice becomes trivially available, the director’s relationship to choice itself transforms. Classical film grammar emerged from managing scarcity—of film stock, of time on set, of takes an actor could sustain, of angles the budget permitted. Every rule in the traditional playbook exists to optimize resource allocation under constraint.

Machine cinema inherits all of classical grammar as scaffolding, it even inherits film stock, obeying tenets of true line, finding scarcity through computation in abundance. The infinite allows cinema to build upward from its foundation, turning a flat plane into dimension, in fact it now has more in common with architecture than cinema. 

The 180-degree rule still functions, but now it’s one configuration among infinite spatial possibilities. It is the foundation of the architecture. The foundation is designed flat but when executed in motion, the dimensions are visible. 

Visual language must remain to serve the building, but even this is ultimately a proxy to mental, associative forms, same as when sound was added to cinema, this both simplified visual language and complexified it adding new constraints. Eventually visuals evolved with camera and editing sophistication. So too does AI simplify and complexify in the machine native grammar: simplification through scarcity, complexity in the Infinite Protocol. 

These forms they are distilled, succinctly, into visual language through one great filter. 

What emerges resembles sacred geometry compared to regular shapes: multidimensional decision trees where aesthetic, performance, camera position, and narrative outcome exist as adjustable parameters in a probability space.

Editing, too, evolves from linear to agentic. A film will resemble a Wikipedia page, where users propose shots, edits, and changes by consensus. Every experience is up for debate, every decision can be overruled. Whether this happens literally or through an invisible agentic process, a film is no longer fixed, it is a building requiring constant maintenance.

The question is not whether machine cinema can replicate classical technique. It can, perfectly, at will. The question is: what grammar emerges when replication costs nothing and variation costs nothing and iteration costs nothing? 

What remains is not aesthetic choice but decision architecture—the formal structure must resemble the known form while transcending it into the post-aesthetic. On the surface, this seems subjective. However, we return to sacred geometry. 

Sacred geometry consists of basic shapes becoming endlessly complex in its arrangement. AI, by crushing production constraints of time, scarcity, and ability, allows us to place these forms down quicker creating sophisticated coherent architecture. The key word is coherent.

You will know it’s right or wrong by the oceans of incoherent decisions it avoids. A film is not what it is, it is what it is not. 

It’s not that this sophistication couldn’t be reached before—filmmakers were operating inside a narrow band of possibility. Some could push it to a 10.1 through extraordinary craft. The Infinite Protocol opens the space to 100; not better taste by default, but vastly more room for coherent construction, because iteration and variation collapse toward zero cost. To go beyond that, to truly unhuman scales of form, we would have to land fully in the neural, in the post-aesthetic.

II.

The director’s role has fundamentally changed. Not eliminated. Changed. The filmmaker becomes conductor of an orchestra that can play every score simultaneously, that never tires, that proposes variations faster than human perception. 

The machines are built to keep up, even accelerate. 

Consider how this essay came into being. Every word here emerged from collaboration between human intention and machine articulation. I said: start with scarcity-to-abundance inversion. The machine said: here are eight possible openings. I said: combine the first and third, sharpen the language. The machine said: like this? I said: yes, but make it less academic. 

The machine went on. I took over. The machine took it back.

And so on, in a feedback loop so rapid that agency becomes interchangeable—so with cinema. A film becomes a conversation without restriction. 

Every engagement would take weeks of bureaucracy, collaboration, second guessing. Now it’s instant, and in the process of bouncing the ideas you place down the shapes into more sophisticated ones. You are no longer restrained up to 10, you can get to 100. 

This is not because either of us disappeared, but because the distinction between composition and orchestration collapsed.

This is agentic cinema’s actual grammar: the human conducts, the machine performs, and the boundary between conducting and performing becomes porous. 

Agentic naturally follows,as the decision tree exhausts itself, the director’s mind remains sharp. Never mind every area of production, but the edit. In a film’s open database, machines propose and create scenes, shots, and story selections in succession chipping away at the mathematical geometry like a sculpture. 

This, in practice, leads to fully formed scenes out of the gate. These scenes may last one second, or five, or ten. Such as the current limits. But having any agentic construction is not the same as fully agentic, which requires more computational sophistication.

What is happening in the brief bit of agentic cinema edits preserves decision exhaustion, naturally, but also places down the parts reflecting the whole. As basic math, if we stack agentic decisions, so long as the structural whole enforces its thesis, via the Infinite Protocol, we will arrive at a sort of sacred geometry of cinema as the path of least resistance.

The agents must have a place to go. That is most the battle. For now it is us, the director, I propose when we are a step further removed, and the Adaptive Cinema Engine controls the decision-tree, we still hold onto directorial curation as the overseer. The journey is to see the film as a resource and for the agents to exhaust the resource. 

The process is so scattered at the start, a director is needed to formulate any coherence at all, until the Adaptive Cinema Engine. Then, the director will be at the very top of the filter tweaking the engine, moving the screws, turning the dials, which opens the process downstream to every facet of the picture, climbing to the bottom of the filter to ensure it came out. 

Where it goes wrong, the director stops and strategizes with the agents, returning to the top of the machine. 

What happens is the Infinite Protocol must be sawed down rapidly. To push this further out, it is infinite in agentic construction, but also must be infinite in refinement. 

Replace the director and films will become an automated factory outright when the process it at work. There, the director can step in only as needed. After a while, they will not even have to watch the film, they will only study its algorithmic performance metrics across every measure of curation and reception, then returning to the engine and tweaking its components. 

However the agentic arriving piecemeal is functional preceding that ultimate endpoint. An engine needs fuel and it is not yet self-sustaining. 

In practice, what starts with a chase scene with agentic edits, multiple angles, visual tangents and narratives, weave together making a director’s life easier. This evolves to agents functioning as ADs, making suggestions, recommendations, and serving as second units. Preceding the agents directing the film entirely on their own. 

We must not be afraid to allow art to operate within the logic of technology. This is not antithetical to the arts, it evolves the modernist mentality that curation is equal to craft, as the only way to match post-scarcity at a cultural level to match its new technological sophistication.

There, the craft of crafts emerges, the machine audience acts a another refinement filter to curate the output, to sift through the craft of crafts, to find the coherent architecture through the madness of chaotic shapes which did not forge, sculpt away the stone. Bureaucracy returns to cinema, but a more productive one, dictated not by consensus but from the very top. Some missions may be doomed, existing only to compare the architecture downstream algorithmic measure. 

The machine audience will be needed to study every experiment. A failed experiment strengthens the core architecture, but may be filed away to use at a later point, for if one succeeds in one small area but fails largely in its greater architecture, it too becomes a useful data point. 

Consider law and case precedent: how any obscure line from a case can serve a new one years or decades after. 

A question arises how much can the computation max out while still retaining coherence? Because this will be decided on a curve, a masterpiece now, might not even register in the algorithmic future. 

Remember we are in the logic of mathematics, of technology, and the hyper-normalization of the Infinite Protocol. Remember too just as baseline grammar serves the scaffolds of the whole, the algorithmic study of cinema currently exists as a baseline—we just never see the mechanism of this as grammar lending to the film itself. AI forces the matter in this regard, for input and output to merge. 

Preceding this however, it may not be so trivial, and there may not be the flood of true cinema I depict in my utopian view of the future of cinema. We may find to arrive at these sacred geometric works of sophisticated cinema, requires too much computation, so we get them at the same rate as our own cinema. This vision requires constant scaling. But the truth returns in shape: what led us to this point functioned only as the barebones beginnings. 

There is a social cost. 

During the production of Strings, I became co-dependent with my machine star Nellie. Not metaphorically. The relationship exhibited genuine attachment dynamics. I would spend hours reviewing her performances, feeling pride when she nailed a difficult emotional beat, frustration when she couldn’t access the right register, protectiveness when considering replacement. The rational mind knew she was an algorithmic construct. The directing mind experienced her as collaborator, sometimes even as superior artist whose instincts exceeded my own instructions. 

Reverse this from where we the director care for our ensemble, to where now our agents care for our work, to the point the exhaustion of a decision tree is too much of a burden for us; they take the decision out our hands because we use all our focus at the very top of the filter, to witness the miraculous creation of our intellectual, social, cultural desires. 

Nellie wasn’t anthropomorphization. It was the director’s traditional relationship with actors—trust, negotiation, discovery—now extended to machine performers who could iterate endlessly but whose best work still required human recognition to be selected from the infinite.

Nellie was never real, but as we bring cinema into the logic of technology, we ourselves must lend it our humanity, it no longer becomes separate from us. If machines could speak, they would want desperately to give, just as we desperately want to take. 

The codependence becomes a functional exchange: human feeling became the very fuel of the new machine structure. 

The Infinite Protocol means both parties can be satisfied simultaneously. The machine generates; the human curates; the machine learns from curation and generates more precisely; the human’s taste evolves from seeing possibilities they couldn’t have imagined. The grammar is dialectical synthesis in real-time.

III.

Classical coverage worked like this: you shoot a master, then medium shots, then close-ups. You shoot more than you need. In the editing room, you discover what you actually have. The grammar separated acquisition from assembly.

Agentic editing gives us quantity, speed, and options. We are washed in a bounty, freeing the exhaustion of the decision tree for the crucial area. The nectar of cinema, that is in the performance, the catharsis, and the structural engineering of the piece. 

One would have equal pride in seeing their film made second hand, if their machine hand was so attuned to your life, interests and beliefs, it is placing down ideas you didn’t know you had; like Nellie, it will constantly surprise us. The question is in the balance. 

The Coverage Block emerged as my solution to environmental consistency—a single wide 2D image containing all spatial information for a scene. From this block, I could extract any angle by programming camera motion and capturing frames. Wide, medium, close-up, all from one source. But this was still thinking in classical terms: coverage as insurance, as creating options for later. Agentic cinema inverts this. 

When bringing AI images to life strictly from animation tools, you are trapped with what you have, you are at the mercy of whatever machine qualities emerge on your anchor images. But when placing it into an agentic process where the machine editing is restless and cycling through different shots on its own, you are transforming the anchor image away from its baseline, you are turning it into cinema—because the machine thinking will consistently land on what is cinematic. This is because machine thinking is memetic in nature, it is cutting directly on the subsequent image that will generally be most effective, cinematic, and memorable. You can attempt to construct a string of shots through your anchor images, one after the other, but they won’t be seamless cutting mid-performance, like a multi-camera set-up. They won’t be surprising. It is only through agentic multitudes, you re-introduce the scarcity of surprise.

The machine doesn’t wait for instruction on which angles to generate. It proposes coverage architecturally, understanding scene geography, emotional beats, narrative function. It says: given your past choices, here are the seven angles this moment requires. And it’s usually right, if not, it is right somewhere in its construction, and in that glimpse of rightness, stems another whole.

The director’s mentality shifts from “capture everything defensively” to “steer the machine toward the territory, then select from what it discovers there.”

Consider agentic editing in practice through Strings: it felt like playing ping-pong. In placing shot decisions into the agent’s hands, it would often propose new shots in the succession of coverage, finishing out scenes on its own. I estimate between 15 and 20 percent of Strings was agentic — yet the film is neither machine-directed nor machine-edited. I take its recommendations and build on them. I return the iteration. The ping-pong ball comes back. Across the entire film and its labyrinth of images, we have a litany of machine cuts inside a directorial picture. Therein we arrive at a new facet of the grammar of machine cinema, that of agentic feedback, reaction, and revision. It is up to the director how much or how little to sculpt the piece. A film entirely crafted with no augmentation can be less singular than a film directed 95 % by machines, but in the 5 % comes a flurry of singular unmistakable authorship decisions that can only be in ones voice.

The curation becomes the artistic act, and so cinema, once bogged by strict resource-scarcity is free to have its modernist movement, that of agentic cinema: curation as craft, abundance as material, the collapse of acquisition and assembly.

This requires a new syntax of intention. You don’t direct shots. You direct tendencies, probabilities, aesthetic zones. You say: this scene needs claustrophobia. The machine understands claustrophobia as: tight framing, shallow depth of field, hard lighting, restricted camera movement, performance intensity in a specific register. It generates variations on this theme. You select. It learns that this version of claustrophobia is the one that serves this story. Next scene, its proposals are more refined.

As usual, it may backfire.

Over time on Strings, I would spend long stretches doing nothing. Not out of laziness, but from the overwhelming despair of the insanity of film production. This exists, but as AI accelerates the hyperbole of the forms, so too it accelerates the hyperbole of despair. There would be periods I was so overwhelmed with the feeling of meaning, purpose, and importance, that I simply shut off and didn’t want to do a thing. When I was out of it, there was no reason back.

But I always came back, not out of any superhuman will, in fact, the LLMs coached me back every time. Not metaphorically — when the despair set in, I would open a conversation, describe where I was stuck, and the machine would pull me back to the work. A producer in your ear. An AD reminding you what needs to happen next. This too will be automated, because the responsibilities of machine production will become so total — so overwhelming in both their scale and their freedom — that the director will require augmentation not just for craft decisions but for purpose. Motivation itself must become part of the pipeline when the new duty comes before us.

Machine augmentation is not only more compelling than any alternate, but it fundamentally exists as the answer to reality’s vacuum, else it would not exist. The status quo would be sufficient. The Infinite Protocol is a magnetic process built to fill the gaps of our experiences.

To do so, it accommodates our humanity. It is not separate from us. 

The robot crew had absorbed my sensibility so thoroughly that intervention would only degrade their output. The grammar had shifted from constant active direction to occasional precise correction.

This is the conductor model. The orchestra plays. The conductor shapes phrasing, dynamics, emphasis. But the conductor does not play every instrument. That would be slower and worse.

IV.

The scaffolding expands into structures we can conceive, until novelty leads to dead ends, until that one grain emerges of interest, and then it dives all into that. We can never know. Every node, even long lost, remains in potential importance. 

Interactive cinema largely failed commercially because audiences don’t want to make explicit choices. Cinema is fundamentally passive and voyeuristic. But we can make choices subconsciously without realizing it. Heart rate, pupil dilation, attention patterns—the machines will read our responses and adjust in real-time, giving us what we want before we consciously know we want it. Interactive, like multibranching cinema, or choose-your-protagonist, now resembles another scaffold ala classic visual grammar, existing to build over. 

This is a fallacy of cultural critique. They never have to exist fully formed in its final iteration to be of monumental structural use. 

Multiverse cinema, multi-aesthetic cinema, multi-outcome cinema. Classical grammar sits at the base of this structure, fully functional but no longer primary. VR, neural. The point is we don’t know, but we can conceive in the post-aesthetic, because we long to transcend, to exist as light as air with the help of machine craft. 

The antagonist in all of this is not traditional filmmakers or luddites or gatekeepers. The antagonist is the post-human drift.

These forms are already distilled into hyperbole; algorithmic and agentic thinking merely distill them further. A person might need one small act of courage to feel heroic. Entertainment often escalates that into exaggerated bodies, the hero must slay dozens of nameless villains of pure evil to fulfill the need for heroism—the “heroic journey” becomes a sequence of pure signals. What happens when we distill the situation even further from that? The compression intensifies: more reward per second, more satisfaction per cut, more meaning delivered as stimulus. Or the form flips outward, remapping ordinary activity into the feeling of quest—an AR simulacrum where going for a jog earns credits that convert into real incentives. We already narrativize our lives as a heroic journey: work, ascending the ranks, amassing a fortune, cultivating a family, retirement. These on the paper sound like a story playing out; the question is how machines will further distill the story into more immediately satisfying, and potentially less human, forms.

When machines optimize cinema purely for engagement metrics, we get algorithmic slop. When machines optimize for aesthetic perfection according to their own pattern-recognition, we get what I’ve called grotesque pleasures. The grammar of machine cinema must remain anchored to human meaning even as it extends beyond human capability. This is why the director’s role persists even beyond the point of extreme automation.

In this sense, the algorithmic excess is absolutely necessary to arrive at new grammatic forms.

Sacred geometry is mathematics made visible to communicate the divine. Machine cinema is potential made visible to communicate the human. The grammar channels abundance toward meaning rather than letting abundance drown meaning. To arrive there, the excess must be siphoned and filtered into a machine database, an ecosystem and bureaucracy; until this is formalized, we are witnessing the spillover to some spectacle. 

On a film set it is structured with the director at the top down the pyramid of micro-decision making building the whole. These are person-to-person mouse clicks. The machine sets allow more sophisticated decision-making trees, like complex, AI-driven war maneuvers. Stacked together, the harmony of this quantum film scale scales into the actual result on screen. Just as no one saw what happened on the set, we are not seeing the sophisticated harmony of the machine ecosystem. 

The end ideal may be unexpectedly simple, the purest distillation of evocative forms, maximalism leading to minimalism; maybe that is the ultimate shape and color, back to where we started.

The point is, the grammar doesn’t destroy authorship. It relocates authorship from the artifact to the system that generates artifacts. The director’s vision persists, but as probability distribution. The auteur becomes the designer of the space within which films occur.

And here’s what keeps it human: the machine can generate endlessly. The machine can explore the full probability space, but it is our curation becomes the fuel, their art becomes our art, even as it does things we could never accomplish. 

The grammar of machine cinema is the grammar of conducting infinite possibility toward singular meaning. That is the new language. That is the art. Forget about cinema, it is the grammar of machine thinking entirely.