A filmmaker’s journey from Soviet theory to AI practice

I remember the first time I encountered Sergei Eisenstein’s “Battleship Potemkin” as a film student. The Odessa Steps sequence wasn’t just editing—it was philosophy in motion, ideas colliding like atoms in a particle accelerator. Fast-forward to 2025, and I’m sitting in front of my computer, wondering: can Google’s Veo 3 understand the dialectical genius that revolutionized cinema nearly a century ago?

The answer, as I discovered through weeks of experimentation, is both simpler and more complex than you might expect.

The Theory Behind the Revolution

Before diving into prompting techniques, we need to understand what made Eisenstein’s intellectual montage so revolutionary. Eisenstein argued that “Montage is conflict” (dialectical) where new ideas emerge from the collision of the montage sequence (synthesis) and where the new emerging ideas are not innate in any of the images.

Eisenstein’s theory towards arts, more specifically filmmaking, is the presentation and construction of CONFLICT. In his theory of dialectic editing, new concepts are generated through the juxtaposition of opposing individual elements and ideas (Thesis+Antithesis=Synthesis).

Think of it this way: show a worker, then show a machine, then show gears turning—suddenly, the audience doesn’t just see three separate images, but comprehends the entire industrial system and the worker’s place within it. The meaning emerges from the collision, not from any single shot.

The Challenge: Teaching AI to Think Dialectically

When I first attempted to create intellectual montage with Veo 3, my instinct was to describe each shot separately:

“A close-up of weathered hands” “Industrial machinery churning” “Steam rising from factories”

The results were visually compelling but conceptually flat. Veo 3 can generate synchronized dialogue, ambient sounds, and background music, producing clips that feel remarkably lifelike, but I wasn’t leveraging its true potential for conceptual storytelling.

The breakthrough came when I realized I needed to prompt for the collision itself, not just the individual elements.

The Veo 3 Intellectual Montage Framework

Structure 1: The Dialectical Trinity Prompt

Aim for a balanced range: roughly 3–6 sentences, or 100–150 words. This gives you room to describe the subject, context, action, and style. Here’s the framework I developed:

[THESIS ELEMENT] cuts sharply to [ANTITHESIS ELEMENT], creating immediate visual conflict. The juxtaposition reveals [SYNTHESIS CONCEPT]. Soviet montage style, high contrast black and white, dramatic angular lighting. Quick cuts between opposing forces generate new meaning through collision rather than narrative continuity.

Example in practice:

Extreme close-up of a golden pocket watch ticking methodically cuts sharply to homeless person's empty hands reaching for scraps, creating immediate visual conflict between wealth and poverty. The juxtaposition reveals the artificial nature of time as social construct. Soviet montage style, high contrast black and white, dramatic angular lighting. Quick cuts between mechanical precision and human desperation generate new meaning about economic inequality through collision rather than narrative continuity.

Structure 2: The Conceptual Collision Prompt

This is a fantastic technique for creating dramatic tension and showcasing a character’s complexity or volatility. You explicitly tell the AI to begin the clip with one specific emotion or state of being and to end it with a completely different one.

Begin with [CONCRETE IMAGE], transition through violent edit to [ABSTRACT CONCEPT VISUALIZED], ending with synthesis of [NEW UNDERSTANDING]. Eisenstein-style intellectual montage emphasizing ideological conflict. Sharp cuts, no smooth transitions, meaning emerges from collision of opposites.

Example:

Begin with factory worker's calloused hands operating machinery, transition through violent edit to golden coins falling in slow motion, ending with synthesis of hands and coins morphing into prison bars. Eisenstein-style intellectual montage emphasizing class struggle. Sharp cuts, no smooth transitions, meaning emerges from collision between labor and capital.

Structure 3: The Metaphorical Warfare Prompt

This technique treats concepts as warring forces within the frame:

[CONCEPT A] and [CONCEPT B] engage in visual combat through rapid montage. [SPECIFIC IMAGERY A] advances, then [SPECIFIC IMAGERY B] counterattacks. Resolution through [SYNTHESIS IMAGERY]. 1920s Soviet cinema aesthetic, stark contrasts, propaganda poster lighting.

Advanced Techniques: The Kuleshov-Veo Hybrid

Building on the foundation of Soviet montage theory, I discovered that Veo 3 responds exceptionally well to emotional direction combined with intellectual concepts:

Juxtapose mother's loving gaze with war memorial statue, then child's innocent laughter with archived footage aesthetic. Each cut intensifies emotional conflict between personal love and collective sacrifice. The collision generates understanding about the cost of peace. Sergei Eisenstein intellectual montage style, emotional crescendo through dialectical editing.

The key insight here is that “The essence of cinema,” Eisenstein wrote, “does not lie in the images, but in the relation between the images!”

Common Pitfalls and Solutions

Pitfall 1: Over-explaining the Concept

Wrong: “Show capitalism and communism fighting and then show how they create a new economic system” Right:“Corporate boardroom handshakes cut to empty factory floor, collision generates understanding about economic displacement”

Pitfall 2: Ignoring Visual Conflict

Wrong: “Peaceful transition from nature to city” Right: “Violent edit from pristine forest to industrial wasteland, no transitional elements”

Pitfall 3: Forgetting the Synthesis

Remember: According to prominent Soviet director Sergei Eisenstein, there are five different types within Soviet Montage Theory: Metric, Rhythmic, Tonal, Overtonal and Intellectual. Intellectual montage specifically requires that third element—the new idea born from collision.

The Future of Dialectical AI

What excites me most about this exploration is how it reveals AI’s potential for philosophical filmmaking. We’re not just creating prettier visuals—we’re teaching machines to think in metaphors, to understand that meaning can emerge from conflict.

The year 2025 marks a pivotal moment in the history of cinema. For the first time in human history, individuals can create Hollywood-quality films that engage with complex philosophical concepts using techniques pioneered by cinema’s greatest theorists.

Practical Exercise: Your First Intellectual Montage

Try this prompt structure and modify it for your concept:

[Choose a social issue] manifests as [concrete visual A] cutting violently to [opposing concrete visual B]. The collision births understanding of [deeper truth]. Soviet montage aesthetic, high contrast, sharp angular cuts, meaning emerges from visual conflict not explanation. Eisenstein intellectual montage principles applied to contemporary issue.

The beauty of Eisenstein’s intellectual montage was its belief that audiences were intelligent enough to complete the thought. Veo 3, when properly prompted, can create those same spaces for intellectual discovery.

As I continue exploring this intersection of century-old theory and cutting-edge AI, I’m convinced we’re witnessing the birth of a new cinematic language—one where the collision between human philosophy and machine capability generates entirely new forms of meaning.

What concepts will you bring into collision next?

About the Author: An AI filmmaker exploring the intersection of classical film theory and generative technology. Follow for more experiments in computational cinema.

Next Week: “Prompting Hitchcockian Suspense: How Veo 3 Learned the Master’s Timing”