-
-
'war1939.pl prequel' series
Inspiration
war1939.pl prequel (originally wojna1939.pl prequel) is an epic military action-thriller based on the novels of bestselling Polish author, Marcin Ciszewski. The series reimagines the events leading to 1939 through the lens of time travel and modern warfare - brought to life with a pioneering blend of next-generation AI tools and human storytelling.
The starting point was a challenge long considered impossible in the Polish market: How do you film a sweeping, high-budget, multi-location military epic with limited resources?
Ciszewski’s novels - full of tactical detail, shifting timelines, and blockbuster-scale scenes - waited years for a screen adaptation because traditional production methods made them cost-prohibitive. This became the project’s core inspiration: to break the long-standing barrier between imagination and feasibility and prove that ambitious genre stories can be told without Hollywood-scale budgets.
What it does
The series reimagines the events leading up to 1939 through a blend of alternate history, time travel and modern warfare, told across 12 episodes. It delivers:
- high-intensity action sequences traditionally too costly to stage
- rich military detail and historical world-building
- believable human characters crafted through AI visuals
- production pipeline that merges classical filmmaking disciplines with cutting-edge tools At its core, it is a military action-thriller shaped by AI but directed and emotionally built by humans.
How we built it
The series is powered by AI as well as constructed using full film grammar:
- a director (Jaroslaw Zamojda) supervised the creative vision, blocking, staging and shotlist.
- the team built detailed scenography catalogues, castings for AI-driven characters and prop libraries (from accurate uniforms to helicopter variants like the Black Hawk).
- every shot required multiple “takes” - 15 to 20 iterations - mirroring traditional filmmaking.
- AI-driven characters were created through a casting-like process to ensure consistency of facial features, body proportions, emotion ranges and wardrobe.
- the team built extensive catalogs of locations, uniforms, weapons, helicopters (e.g., accurate Black Hawk variants), vehicles, and interiors.
Challenges we ran into:
- Complex multi-character scenes: AI struggled to place more than 3–4 consistent characters in a single frame; specialized workarounds were developed to achieve complex group shots.
- Rapidly changing tools: New models appeared daily, making early episodes appear technically outdated within weeks - forcing constant recalibration and a unified visual language.
- Logistics of remote collaboration: A dozen artists, each highly creative and independent, had to synchronize style, pacing and technical standards.
- Precision in historical and military detail: Aircraft, vehicles, uniforms, insignia, landscape and architecture had to be historically accurate and consistent with the writer's universe.
- Continuity & logistics: Maintaining consistent character appearance, wardrobe, lighting and camera logic across a remote multi-city creative team required film-level coordination.
- Emotional realism: Achieving believable facial expressions and nuanced emotions demanded manual correction, direction and dozens of iterative passes.
Accomplishments that we're proud of:
- Created one of the first large-scale AI-driven action series in Poland developed with a full filmmaking workflow.
- Achieved cinematic sequences - White House galas, helicopter operations, large tactical deployments- that would have been financially impossible in standard production.
- Developed custom techniques for multi-character AI scenes, continuity preservation and realistic emotional expression.
- Built a 12-episode narrative world that preserves the tone, scale and tension of Ciszewski’s novels.
- Proved that a Polish production can operate at a visual scale previously reserved for Hollywood budgets.
- Established a pioneering human + AI production pipeline used as a blueprint for future projects.
What we learned
AI filmmaking is not easier - just different. Many scenes that appear effortless on-screen (like a 100-person gala in the White House) required complex coordination, custom techniques and persistence, because AI often resisted placing more than three characters in a frame consistently.
Coordination becomes its own art form. The team operated like remote film crew dispersed across multiple cities, learning to unify style, continuity, lighting and character likenesses across thousands of generated frames.
Technology evolves faster than production. Tools improved daily, meaning shots created two months earlier already felt outdated. Maintaining visual consistency across 12 episodes, while technology kept reinventing itself, was one of the project’s biggest insights and challenges.
AI requires creativity, not shortcuts. The most successful moments came from imaginative problem-solving - designing unusual workflows, “outsmarting” the AI when it fought back and inventing new techniques the industry hadn’t yet formalized.
Human emotion remains irreplaceable. Conveying convincing facial expressions and emotional nuance required manual control, direction and refinement - AI alone at once could not deliver believable human performance.
The next stage of the project includes:
- completing and releasing all 12 episodes
- expanding the universe toward a full adaptation of Ciszewski’s novel WWW.39
- iterating on the production pipeline to take advantage of the newest AI generation tools.
- preparing international festival submissions, beginning with the Chroma Awards.
- developing behind-the-scenes materials to share our methods and support other creators exploring hybrid AI filmmaking.
Ultimately, this 12-episode prequel is just the beginning. lt became not just a series, but a proof-of-concept for a new cinematic pipeline - one where AI expands creative possibilities, but human vision still leads every step. It’s a project built at the frontier of filmmaking, by a team determined to show that bold, large-scale stories can now emerge from places where they were previously unlikely to produce.
Log in or sign up for Devpost to join the conversation.