YOUNG COP

1080p Is the New Baseline. Here's What That Means for the Film.

When I started production design in May, Higgsfield Cinema Studio generated at 720p. That was the ceiling. Every frame of the film was planned around a 720p-to-4K upscale pipeline using Topaz Video AI. The math was simple: generate at 720p, upscale 4x, deliver at 4K.

That math just changed. Higgsfield Cinema Studio 2.0 and now 3.0 output at up to 4K natively. But more importantly for this project, 1080p generation is now fast, cheap, and visually superior to what 720p looked like even two months ago. The cost per minute of AI video generation has dropped roughly 40% since I started.

What does 1080p source material mean for Topaz upscaling? Everything. Topaz Video AI reconstructs detail from what it has. Give it 720p and it has to invent a lot. Give it 1080p and it has 2.25x more pixel data to work with. The 4K output from a 1080p source is sharper, more detailed, and has fewer artifacts than 4K from 720p. Skin texture holds. Text is legible. Fabric weave is visible. It is a meaningful, visible improvement.

The practical impact: I am regenerating key scenes at 1080p for the teaser trailer. The full film production starting in late July will generate entirely at 1080p. The Topaz upscale to 4K will be cleaner, faster, and require fewer passes. Estimated cost savings on compute: 15-20%, because fewer frames need re-generation for quality issues.

This is the curve I talked about in earlier posts. Every month the tools get better and cheaper. The teaser trailer on July 4 will be pure 1080p source upscaled to 4K. That is the final pipeline. By the time the full film hits screens in November, the quality gap between this and a traditionally shot indie will be smaller than anyone expects.

Total estimated cost change for the 4K pipeline: down from $450-$650 to approximately $350-$500. Same quality. Actually, better quality. Less money. That is the trend line. And it is not slowing down.

Adobe Just Bought Topaz Labs. Here's What That Means for Young Cop.

Today Adobe announced a definitive agreement to acquire Topaz Labs, the AI video and image enhancement company whose tools are a core part of this film's production pipeline.

Topaz Video AI is how Young Cop gets from 720p (the native output of Higgsfield Cinema Studio) to 4K delivery. Every frame of the finished film will pass through Topaz's upscaling models before final export. It's not optional. It's load-bearing infrastructure.

So what does an Adobe acquisition mean for this project? In the short term: nothing changes. Topaz's standalone tools will continue to work as they do today. The deal isn't expected to close until the second half of 2026, and CEO Eric Yang is staying on to lead the team. My production timeline (trailer July 4, full film November 20) fits well within that window.

In the longer term: this is validation. Adobe doesn't acquire tools that don't work. Topaz's AI enhancement models are now considered essential infrastructure for professional creative work. That's exactly how I've been using them. The tool I chose for upscaling an indie AI film just got bought by the largest creative software company on the planet.

There's also a practical upside. If Topaz's models eventually integrate into Adobe's Creative Cloud or DaVinci Resolve plugins, the workflow gets simpler. Right now I'm running Topaz on cloud GPUs through Vast.ai. If that becomes a one-click operation inside the editing suite, future projects get faster and cheaper.

For now: the pipeline holds. Topaz Video AI remains the tool. The 4K upscale strategy is unchanged. And the fact that Adobe validated this specific technology on the same day I'm building a film with it feels like confirmation that the approach is sound.

The Script Is the Real Version. The Novel Was the Sketch.

I want to be honest about something. The novel was first. It was necessary. But the screenplay is the version of this story that actually lives.

The novel was me figuring out who these people are. Writing their internal monologues, living inside Caleb's head, exploring every corner of the town at my own pace. It served its purpose. It let me find the story without a camera watching.

The script is what the story became once I stopped explaining it and started showing it. Every scene was earned. Every line of dialogue was pressure-tested. If a character's interior thought couldn't become visible behavior, it got cut. If a V.O. line told the audience something the image already showed, it got cut. What remained felt more alive to me than the novel.

Specific things the script does that the novel doesn't: Marcus's silence when his tapping stops. The doodle on the notepad during the stakeout. Coleman bowing her head over French toast. Erin's pen cap snapping in two. These are not in the book. They emerged in the adaptation because a screenplay forces you to make interior life into exterior action. That process made the story better.

The novel is free. Read it. It's the foundation. But if you want to know what this story really is at its best, the film is the answer. The script is 79 scenes of a story that has been rewritten, pressure-tested, and refined until I believe every beat earns its place. The novel was the first draft of the soul. The script is the finished house.

And the film will rise above the script the same way the script rose above the novel. A script describes a silence. The film is the silence. A script says the rain falls on tin. The film is the sound of it. A script notes that Marcus taps a rhythm on the door. The film is that rhythm in your chest.

Each form elevates the last. The novel found the people. The script gave them bones. The film gives them breath. By the time you hear a fourteen-year-old voice crack on the radio, or see Coleman's face when she bows her head, or feel the exact weight of three seconds of silence between Reno and Caleb in the mud, you will not be reading a story. You will be inside one.

We're Months Away from Story-to-Screen for Under $100. What Does That Mean?

I believe we are a matter of months away from being able to go from story to script to multiple distribution channels at a cost of under $100. And perhaps in the future, under $1.

Think about that for a moment. A complete narrative - written, scored, voiced, animated, edited, and distributed - for less than a nice dinner. The tools are converging fast. Video generation improves every month. Voice synthesis is already indistinguishable. Music generation from a simple prompt or a piano melody is here today. Editing software is free. Distribution is a URL.

What does this mean for Hollywood? For the studio system? For the thousands of people whose livelihoods depend on the current model of production? These are real questions and they deserve honest answers, not hype.

There is legitimate concern about quality. When the barrier to creation drops to zero, the volume of content explodes. Most of it will be bad. Most of it will be derivative, rushed, or soulless. That is a real risk and it would be dishonest to pretend otherwise.

But in the end, audiences will decide. They always have. They chose television over radio. They chose streaming over cable. They chose YouTube over network programming. Not because those things were cheaper to make - but because the best of them were worth watching. The cream rises. It always has.

The question is not whether AI-generated film will replace traditional production. The question is whether a single person with a story, a laptop, and $100 can make something that moves an audience the way a $200 million production does. I believe the answer is yes - not because the technology is magic, but because storytelling is human. The tools change. The craft does not.

We are about to find out.

Building the Trailer, Planning the Kickstarter: What Comes Next

Right now I'm deep in trailer production. The goal is a full-length teaser trailer that demonstrates the story and the artistic vision using the full technology workflow: Higgsfield for video generation, ElevenLabs for voice, Suno for music, DaVinci Resolve for editing, real sound effects for grounding. A trailer that shows what this film looks and feels like when all the pieces come together. Targeting July 4.

If the trailer proves the concept the way I believe it will, I'm taking Young Cop to Kickstarter. Not to fund distribution - that's Amazon, and it's free. The Kickstarter funds production: the AI compute, voice generation, cloud rendering, and upscaling that turn a script into a finished film. The total budget is under $7,000. The campaign covers the remaining costs I haven't already spent out of pocket.

Planned tiers: Executive Producer credit in the finished film. Associate Producer credit. Special Thanks in the credits. A Name a Character tier where backers can name a non-protagonist, non-antagonist character in the film. Names must be realistic and appropriate, and I reserve approval. And a Background Appearance tier where backers can appear visually in the film as a background character. No speaking parts, but you'll be in the movie. It's a way to put a piece of yourself into the story permanently.

The film releases on Amazon Prime Video Direct on November 20. The Kickstarter is how it gets made. Amazon is where it lives.

One more thing worth noting: by late July, when full film production begins, the underlying models will have improved again. Every month, the quality goes up and the cost per generation goes down. I'm timing production to ride that curve. The film I make in August will look better and cost less per minute than anything I could produce today.

More details once the trailer is finished and I can show rather than tell. For now: the work continues every night.

Editing Young Cop in DaVinci Resolve: Why Free Tools Are Enough for a Feature Film

The edit is happening in DaVinci Resolve. The free version. No subscription, no watermarks, no limitations that matter for this project. It handles everything I need: multi-track timeline, audio mixing, color grading, and export at full quality.

The workflow is block-based. Each scene comes out of Higgsfield as a defined clip. I'm assembling these blocks on the timeline, then layering in the real sound effects from ZapSplat, the voice performances from ElevenLabs, and the original soundtrack. Every scene gets its own sound bed: footsteps on wet pavement, the cruiser's engine idling, cicadas in the distance, a screen door closing down the street. All of this on an M3 MacBook Air. No dedicated editing rig. No external GPU. A laptop on a desk.

The voiceovers are placed scene by scene. Caleb's narration sits on its own track. Dialogue goes on another. Music on a third. This makes it easy to adjust timing and volume without re-rendering anything. Resolve handles it all in real time, even on my machine.

The key insight is that AI-generated video comes out as finished blocks. There's no raw footage to sift through, no B-roll to sort, no dailies. Every clip is already the final take. That changes the edit fundamentally. Instead of finding the story in the footage, you're assembling a story you already have. The edit becomes about rhythm, pacing, and layering sound. Resolve is built for exactly that kind of precision work.

One more thing: using Resolve today, on a real project at this scale, creates a dated record of the production process. The project files, the export logs, the timeline metadata. It all timestamps the creative work as it happens. That matters for copyright, for proof of authorship, and for showing that a human made the editorial decisions throughout.

Total cost for the editing software: $0.

Production Design Is Done. Every Location, Character, and Prop Is Built.

The full production design for Young Cop is complete. Every location. Every character. Every prop. You can see all of it now under Behind the Scenes on the site.

The visual language of this film is rooted in 1985. Not nostalgic 1985. Real 1985. Faded paint on clapboard houses. Fluorescent light in a convenience store at midnight. A police cruiser with a bench seat and no computer. Rotary phones on kitchen walls. Wood paneling. Screen doors that don't close all the way.

The locations are specific. This isn't a generic Southern town. It's Princeton, Kentucky in the mid-eighties. The diner with its neon bleeding across wet pavement. Coleman's store with the bell on the door. The school behind chain-link. Elm Street with its dark porches. The state road where the cruiser idles. Each location was designed to feel lived-in, worn down, and real.

The characters were designed with the same specificity. Caleb's uniform is two sizes too big. Marcus has a quiet weight to him. Reno looks like someone you'd cross the street to avoid. Miss Coleman looks like she's been running that store since before the town started dying. Every character design tells you something about who they are before they speak.

The props matter too. Caleb's notepad. The badge he touches like a heartbeat. The phone book he sits on to see over the dash. These are small details but they ground the world. They make it tactile.

All of this is pre-visualization concept art. It represents the production design, not the final film. The teaser trailer on July 4 will show the actual visual style, animation quality, and final look. But the design work is done. The world is built. Now it gets made.

Marcus Robinson: The Character I'm Happiest With

This is my first screenplay. I have never written one before. So when I say Marcus Robinson is the character I'm happiest with, that carries a specific weight. He's the one who surprised me. The one who showed up as a supporting role and became the moral center of the whole film.

He's Caleb's partner. Rides passenger. Taps a rhythm on the car door that never stops - until it does. He made a promise to a dead man to look out for his son, and he's been half-assing it for two years. The film is partly about him deciding to stop.

What makes Marcus work: he's not a mentor in the traditional sense. He doesn't teach Caleb how to be a cop. He teaches him what a gun actually means. He teaches him that freezing is better than shooting. He teaches him, eventually, that letting a child do a man's job was the wrong call - even if it saved the town.

His rhythm - the tapping - is his heartbeat on screen. When it stops, something real is about to be said. When it comes back softer, a decision has been made. It's a musical motif performed by a hand on a car door.

By the end of the film, Marcus says something on a bench at dawn that I think is the line that surprised me most: 'I let a child do a man's job because it was easier than saying no.' That's not a mentor speaking. That's a man reckoning with his own complicity in something he should have stopped. And the fact that he says it with love doesn't make it less damning.

For a first script, having one character land like that feels like enough. Everything else is extra.

Five People You'll Meet in Princeton, Kentucky

Caleb Harlen, 14. Orphaned. Drives the night patrol cruiser on a phone book for height. Carries a revolver he was told never to use and a folder he can't put down. Eats cold soup standing at the counter. Prays behind a tobacco barn when no one is looking. The uniform is two sizes too big. The job fits perfectly.

Marcus Robinson, 34. Caleb's partner. Taps a rhythm on the car door that keeps time for both of them. Made a promise to Caleb's father the night before he died. Been on the night shift for twelve years. Funny how that works.

Miss Coleman, late 20s. Runs the general store. Scanner under the counter. Keeps a light on every night. Has been praying over a boy she can't legally keep - yet. The adoption papers are six weeks away. She already bought the sheets for the spare room.

Erin, 14. Smart. Opinionated. Paperback always in hand. Wants to be a writer someday. Gave Caleb The Outsiders inscribed 'Stay gold.' She doesn't patrol with him. She does something harder: she stays.

Bobby McClure. Coaches Little League. Goes to First Baptist. Bought a new truck with cash. The kind of man nobody suspects. The kind who moves the line every month and tells himself there's still a line.

Why the Villain Has a Kid He Can't See

Reno is the antagonist. Leather jacket. Trans Am. California plates. He threatens a fourteen-year-old boy, leaves a Polaroid on his car, and stands on his porch unscrewing the light bulb.

He also has a son he calls from a gas station payphone at 2 AM and never gets to talk to.

I made that choice early. Not to soften Reno. Not to make him sympathetic. But because the film is about what adults do to children - and Reno is doing it to two of them. One he's threatening. One he's abandoning. The difference is a matter of geography, not morality.

In the raid, when the cuffs go on, he yells about his kid. That's not redemption. That's the moment you realize: even the monster has someone waiting at home who doesn't understand why daddy stopped calling.

The film doesn't ask you to forgive Reno. It asks you to see that the damage radiates outward. Every arrest has a child on the other end of it. The audience in a theater may not think about that. The audience in a prison will.

The Town Is a Character: Why Princeton Matters

I wanted to create a world of yesterday to show a timeless problem. The drugs change names. The economics shift. But the pattern - a community bleeding quietly while the people who could stop it choose not to look - that is not 1985. That is always.

Princeton, Kentucky is specific because specificity is what makes a story feel real. Caldwell County. The Black Patch. Tobacco country. A real town with a real First Baptist Church on West Main Street and a real Route 91 running through downtown. I could have invented a place. I chose a real one because real places carry weight that fiction cannot manufacture.

In 1985, the tobacco money was drying up. Allotments shrinking. Price supports barely holding. Young men leaving because there was nothing to inherit. That is the soil the drug operation grows in. People with skills, trucks, and empty barns - looking for cash. That is not a Kentucky problem. That is an everywhere problem dressed in a Kentucky accent.

The film lives in this town the way you live in a place you know by smell. Wet asphalt. Curing tobacco from the barns outside town. Cold coffee in a thermos at two AM. The dogs that know the sound of the cruiser. These details are not decoration. They are the argument. If you can feel the town, you can feel what it costs to protect it.

The novel introduced Princeton. The film inhabits it. The difference is sensory. You will hear the rain on the tin roof. You will see the neon bleed across wet pavement. You will know this place the way Caleb knows it - by what it sounds like at midnight when nothing is open and everything is watching.

The Script Is Done. It's a Different Animal Than the Novel.

The screenplay is finished. Final draft. 80 minutes of story, locked and ready for production.

It's different from the novel in some ways, and that's the point. A script and a book are different vehicles for the same story. They don't carry it the same way. They shouldn't.

A novel lives in a character's head. You get to hear what Caleb thinks when he sees the broken window. You feel the weight of his silence. You know what the rain smells like to him. A script can't do that. A script lives in the cut. It's what the camera sees, what comes out of someone's mouth, and how long the silence lasts between them. Interior life has to become exterior behavior.

Some scenes expanded in the script. A moment that was two paragraphs in the novel became a full scene once I could see the staging, the blocking, the way two characters move through a room without speaking. Other scenes compressed. A chapter of internal monologue became a single look between Caleb and Marcus that says everything the prose spelled out.

The structure shifted slightly. Film has different rhythm requirements. You need act breaks in places a novel doesn't. You need visual setups that pay off faster because the audience can't flip back a page. Some subplots that breathe beautifully in prose had to tighten for the screen.

But the soul is the same. Caleb is the same kid. The town is the same town. The moral question at the center hasn't moved an inch. What changes is how it reaches you. The novel whispers it. The film shows it.

Having both documents means the story exists in two complete forms before a single frame is generated. That's the foundation everything else builds on.

The Soundtrack Is Done. Full Orchestra, 1980s Style, Built from a Keyboard.

I'm a songwriter and producer. Getting the music right was important to me. Not as an afterthought, not as filler between scenes. The score needed to carry the same weight as the writing.

I went with a 1980s full orchestra approach. Warm strings, subtle brass, piano that breathes. The kind of scoring you'd hear in a Spielberg or John Hughes film. Not synthetic. Not minimal. Lush, emotional, and patient. Every track started on piano - I played it, recorded it, and then fed it into Suno. Suno interpolated my playing and I built upon that, directing it into full orchestral arrangements.

The themes and motifs came together beautifully. Caleb has a quiet, rising figure that plays when he's alone in the cruiser. Marcus gets a deeper, slower line. The town itself has a theme. When these motifs weave together in the later scenes, you feel the relationships without anyone saying a word. That's what a good score does.

Eight tracks plus a duet pop theme song coming July 24. All generated through Suno from my piano compositions. I composed and directed every piece. The soundtrack is available now on the site, free under Creative Commons.

And there's one more piece coming. On July 24, I'm releasing a pop theme song: My Love Will Bring You Home, a duet. The idea is to bring the movie to an audience through a song that lives outside the film. Something people can find on its own, connect to emotionally, and follow back to the story.

A theme song is a front door. The film is the house.

The Novel Is Done. Here's How I Made Sure the Story Holds Together.

The Young Cop novel is finished. I'll be releasing it for free under Creative Commons on the website soon. The full story of Caleb Harlen, start to finish, in prose.

Here's a summary: a fourteen-year-old orphan drives the night patrol cruiser in a dying Kentucky town. He carries a revolver he was told never to use and a folder he can't put down. When a drug operation starts taking lives, he connects dots no one else wants to see. The closer he gets, the more it costs the people he has left. It's about faith, morality, and what happens when you're the only one who refuses to look away.

Writing a novel is different from writing a screenplay. I didn't have to write the book. I could have gone straight to script. But I wanted to focus on the story in black and white first. No camera angles, no scene headings, no time constraints. Just the characters, the town, the moral weight of every choice. The novel let me live in Caleb's head in a way that a script never could.

The book is more detailed in some ways than a script can be. Internal monologue, the smell of Coleman's store, what the rain sounds like on a tin roof at 2 AM. A script can't carry that. But a script carries things the novel can't: visual rhythm, the exact duration of silence between lines, the way a cut from one face to another tells you everything without a word. They're complementary documents. The novel is the soul. The script is the blueprint.

Next step: converting it to a screenplay. The story structure is locked. The characters are locked. Now it's about translating interior life into exterior action.

One thing I want to talk about: how I made sure the story actually holds together. Writing a novel over weeks, you introduce ideas early that you evolve or abandon later. Characters shift. Motivations deepen. You can lose track of what you promised the reader in chapter two when you're writing chapter twenty. So I used adversarial analysis throughout the process.

I ran the manuscript through Grok, Gemini, and NotebookLM with specific prompts: find contradictions, find dropped threads, find characters who appear once and vanish, find motivations that shift without justification, find tonal inconsistencies. Not to write the story. To stress-test it. To make sure I hadn't introduced concepts out of order or left loose ends as the story evolved to completion.

Each tool caught different things. Grok was aggressive about logic gaps. Gemini found tonal drift. NotebookLM was useful for tracking character arcs across the full manuscript. Using them together, adversarially, made the novel tighter than any single pass of self-editing could have. The story congealed harmoniously because I treated every draft like it needed to survive cross-examination.

I'll be honest: sometimes I died laughing throwing completely out-of-pocket plot turns or character actions at the AI just to see how it would react. That was genuinely fun. But in the end, that back-and-forth also helped. Having the AI approach my material from every angle and give constructive feedback made me think harder about my choices. That said, I didn't budge much, if at all, from my ideal way to approach the story I created. The feedback sharpened my confidence in the decisions I'd already made.

Why Real Sound Effects Matter for an AI-Generated Film

If the visuals are generated, why bother with real sound effects? Because sound is what makes a scene believable. Your eyes might accept an AI-rendered image, but your ears know immediately when something is wrong.

I'm sourcing all sound effects from ZapSplat. Real recordings. Actual rain on pavement. An actual car door closing. A screen door with a spring. Footsteps on gravel. These aren't synthesized or generated. They were recorded in the physical world.

The goal is to anchor the AI visuals in reality through the audio layer. When you hear real cicadas over a generated Kentucky night, it grounds the whole scene. The sound tells your brain: this is a place. The image provides the look. The sound provides the feel.

Every scene gets its own sound bed built from scratch. Ambient layer, spot effects, foley. Cruiser idling. Radio static. A pen clicking. Coffee being poured. These details accumulate. They make the world dense and real in a way the visuals alone can't.

It also differentiates the film from anything that just plays AI video with a music track underneath. That's a slideshow. This is a movie. Movies have sound design.

4K for Under $650: Upscaling with Topaz Video AI in the Cloud

The film renders natively at 720p. That's the resolution Higgsfield Cinema Studio outputs. It looks good at that size, but for a feature film you want 4K. The question was how to get there without spending thousands on hardware.

The answer: Topaz Video AI running on cloud GPUs through Vast.ai. Two RTX 4090 instances running simultaneously. The full 80-minute film upscaled to 4K in approximately 25-30 hours of compute time.

Estimated total cost: $450-$650 including Topaz licensing and cloud compute. Compare that to buying two 4090s ($3,200+) or renting dedicated hardware for weeks.

Topaz doesn't just scale pixels. It reconstructs detail, reduces artifacts, and sharpens in a way that makes the upscaled output look like it was rendered at 4K natively. The 720p source becomes a genuine 4K delivery.

This is the kind of economics that makes indie filmmaking viable at a quality level that used to require studio budgets. A feature film, delivered in 4K, for the cost of a weekend trip.

Higgsfield Cinema Studio: Generating an 80-Minute Film at 720p

Higgsfield Cinema Studio is where the actual footage comes from. Every frame of Young Cop is generated here. The output is 720p video clips that serve as the raw material for the edit.

The process is scene-by-scene. I feed in the production design, the character references, the script directions, and the visual tone. What comes back are finished clips. Not storyboards. Not rough drafts. Finished takes.

I will generate over 2.5 hours of raw footage to produce the final 80-minute film - over 1,000 individual video generations. That ratio matters. Not every generation is usable. Some need to be regenerated for consistency, performance quality, or timing. But the ones that land, land clean.

The 720p resolution is a deliberate trade-off. Higher resolution means longer generation times and higher costs. At 720p, I can iterate quickly, generate multiple takes, and make creative decisions faster. The Topaz upscale to 4K handles the rest in post.

What Higgsfield gives you that other tools don't: consistent character appearance across scenes, camera movement control, and cinematic framing. It thinks like a cinematographer, not an image generator. That's the difference between making a movie and making a series of pictures.

And the platform is still improving. By late July, when full production begins, I expect higher resolution output, better consistency, and faster generation times. The teaser trailer captures where the technology is today. The full film will benefit from where it's headed.

This Film Would Never Get Made. That Is Why I Am Making It Myself.

Let me be clear about something. Young Cop would never be greenlit by a studio. A 14-year-old protagonist who barely speaks. A Southern noir with no stars. A slow burn with no action sequences. A $7,000 budget. A first-time writer with no representation. There is no version of the traditional system that says yes to this.

It is not that the story is bad. It is that the story is uncommercial by every metric the industry uses to decide what gets made. Too quiet. Too slow. Too young a lead. Too regional. Too faith-adjacent for secular buyers. Too secular for faith buyers. It falls between every category that has a marketing playbook.

This is what AI changes. Not the quality of the story. The ability to tell it without permission. I do not need a greenlight. I do not need a budget approval. I do not need someone in a conference room to decide that a 14-year-old orphan in Kentucky is worth the risk. I can make the film myself, at night, for the cost of a used car, and put it in front of anyone who wants to see it.

That does not mean anyone will watch it. It may find no audience. It may live on Amazon with zero views. That is a real possibility and I am at peace with it. The point was never guaranteed success. The point was: the story exists. In full. On screen. With sound and music and silence and performance. That used to require millions of dollars and a committee's approval. Now it requires time, tools, and stubbornness.

There are thousands of scripts like mine sitting in drawers. Stories that are too quiet, too specific, too uncommercial for a system that needs opening weekends. AI does not make those stories better. It makes them possible. That is a different thing. And for the people carrying those stories, it is enough.

Why I'm Making This Myself

I wrote a screenplay. A Southern noir about a fourteen-year-old cop in a dying Kentucky town. It is, I believe, a genuinely good story. In another lifetime, I would take this to a studio and make it the traditional way.

The traditional path for this project would be: find an agent, submit the script, wait months for reads, take meetings, attach a producer, attach talent, find financing, wait two years, hope it gets made. That system works for many people. It did not work for my timeline or my temperament.

I am not going to do that. I am going to make this film myself, using AI production tools, with one person making every creative decision from script to screen. Not because I think AI is better than a crew. It is not. But because the tools now exist for a single person with a clear vision to execute at a quality level that was impossible two years ago. And because the story deserves to exist on its own terms.

I am not going to wait for permission to make something I believe in. The tools exist. The story is ready. The work starts tonight.

As an aside: I previously worked on the business side at a major entertainment company. They of course have right of first refusal, which I am honoring once the script becomes the shooting script.

The Goal Is Amazon. The Audience Might Be One.

I am not waiting for the industry to validate this project before I make it. My goal for release is Amazon. Self-published to their platform the same way an author puts a novel on Kindle. No distributor. No sales agent. No middleman deciding whether the work is worthy of being seen. Just: here is the film. It is available. Watch it or do not.

My goal for release is Amazon. Self-published to their platform the same way an author puts a novel on Kindle. No distributor. No sales agent. No middleman deciding whether the work is worthy of being seen. Just: here is the film. It is available. Watch it or do not.

I have no expectation of a large audience. This might be a film with a creator of one and a viewership of one. That is fine. I did not make it because I expect millions of people to care. I made it because I had a story that would not leave me alone, and now the tools exist to bring it to life without asking anyone for permission or money.

I think many people will do this someday. Not filmmakers. Regular people. People with a story they have been carrying for years, who never had the budget or the crew or the connections. One day soon, the path from idea to finished film will be something anyone can walk. That is not a threat to cinema. It is a gift.

For me, personally, it is a blessing. To be able to take something from my head and put it on a screen with my own hands, on my own time, with no one to answer to except the story itself. That is all I wanted. Everything else is extra.

If someone at Amazon, a distributor, or a festival sees this and thinks it belongs on a bigger stage, I am open to that conversation. But I am not waiting for it.

Building Every Character, Location, and Prop with GPT Image 2

Before a single frame of film is generated, everything needs to be designed. Every character. Every location. Every prop. That's production design, and for Young Cop it all happened in GPT Image 2.

Characters first. Caleb needed to look 14. Not Hollywood 14. Actually 14. Skinny, unfinished, swimming in a uniform that doesn't fit. Marcus needed quiet authority. Reno needed to look dangerous without being a caricature. Miss Coleman needed decades of patience in her face.

Every character got iterated until they felt like a person I'd recognize on the street. Then I locked their reference images. Those references become the basis for everything else: the film generation, the concept art, the promotional materials. Consistency starts here.

Locations followed the same process. The diner. Coleman's store. The cruiser interior. Caleb's house. The school. Each one designed to feel specifically 1985, specifically Kentucky, specifically worn down but lived in.

Props are where the film gets tactile. The badge Caleb touches. His notepad. The phone book he sits on. The rotary phone on the kitchen wall. Each one grounds a scene in physical reality.

The full production design library is 150+ images. All viewable under Behind the Scenes on this site. These aren't the final look of the film. They're the blueprint the film is built from.

Giving Every Character a Unique Voice with ElevenLabs

A film lives or dies on its performances. For Young Cop, every character needs a distinct voice. Not a generic AI voice. A voice with grain, rhythm, and personality. That's ElevenLabs.

Caleb sounds young but measured. He speaks carefully, like someone who learned early that words have weight. Marcus has a low, unhurried voice. Steady. The kind of voice that calms a room without raising itself. Reno is sharp and unpredictable. Miss Coleman is warm but carries decades of tiredness.

Each character got their own custom voice profile. Not cloned from a real person. Designed from scratch to match the character as I hear them. Over 30 characters in the film will have a unique voice. The quality is remarkable. These aren't robotic text-to-speech outputs. They have breath, pacing, emotion.

The dialogue is performed line by line with direction for tone, speed, and emotional register. A whispered confession doesn't sound like a shouted warning. A joke lands differently than a threat. ElevenLabs handles that nuance.

The audiobook narration uses the same system. Caleb's voice narrates the novel. It's the same performance quality as the film dialogue, which means the audiobook and the film feel like they come from the same world. Because they do.

The voice work also gets layered into DaVinci Resolve on its own track. That means I can adjust timing, overlap dialogue with ambient sound, and mix performances against the soundtrack with full control. No compromises.

Film Distribution Strategy: Amazon First, Self-Hosted Backup

Update: The primary distribution target is Amazon Prime Video Direct. The film will be self-published there the same way a novel goes to Kindle. No distributor. No middleman. The self-hosted infrastructure described below serves as backup and for the free companion materials (novel, soundtrack, website).

Young Cop will also be available directly from infrastructure I control. Cloudflare R2 for storage. HLS adaptive streaming for delivery. The full film chunked into 6-second segments at multiple quality levels, served globally through Cloudflare's edge network.

Here's why this works economically: R2 has zero egress fees. That means no matter how many people watch the film, I don't pay for bandwidth. The storage costs pennies per gigabyte per month. The entire film in multiple quality tiers costs maybe $0.15/month to store.

HLS adaptive streaming means the player automatically adjusts quality based on the viewer's connection. Fast internet gets 4K. Slower connection gets 720p. No buffering, no manual quality switching. It just works.

No ads. No tracking pixels. No third-party analytics watching what people watch. The viewer hits play, the film streams from the nearest Cloudflare edge node, and that's it. Clean.

The total hosting cost for distributing an 80-minute feature film to a global audience: under $5/month. Even if it goes viral. Even if a million people watch it. Zero egress means zero surprise bills.

This is what control looks like. Own the file. Own the player. Own the relationship with the audience. The technology exists to do this now for almost nothing. So that's what we're doing.

AI Isn't Copying Film. There Isn't Enough Film to Copy.

I want to address something directly because the conversation around AI and creative work is dominated by a claim that isn't true: that these models are copying existing film, television, and music. That they are sophisticated plagiarism engines. That every frame they produce is stolen from a human artist.

I am not an AI researcher. But I have used these tools every night for months, and I have read enough to understand what they actually do versus what people claim they do.

Here is the math. In the entire history of cinema - 130 years - humanity has produced approximately 500,000 feature films. Total. All languages. All countries. All genres. The entire corpus of television adds maybe another 2 million hours. The entire recorded music catalog across all of history is roughly 200 million tracks.

These numbers sound large until you understand what a modern generative model actually learns. It learns physics. Lighting. How fabric drapes. How smoke diffuses. How a human face moves when it speaks. How gravity works on hair. How perspective recedes. These are not things you learn by memorizing 500,000 movies. These are things you learn by processing billions of images of the physical world and developing a mathematical understanding of how reality behaves.

A model that generates a frame of a boy sitting in a car at night is not copying any specific frame from any specific film. It understands what light does to vinyl at night. What dashboard glow looks like on skin. How rain distorts a windshield. It learned these things the same way a painter learns them - by looking at the world - except it looked at more of it, faster, and remembered all of it.

The claim that AI is plagiarism requires you to believe that understanding how light works is theft. That knowing what a face looks like is copyright infringement. That generating an image of a Kentucky town at night is stealing from every photographer who ever shot a Kentucky town at night. That is not how knowledge works. That is not how creativity works. And deep down, the people making that claim know it.

What AI actually gives a creative person is leverage. I can write a scene and see it. I can hear a character speak the line I wrote and adjust the delivery. I can compose on a piano and hear an orchestra play it back. These are not replacements for human creativity. They are amplifiers of it. The story is still mine. The characters are still mine. The moral weight of every choice is still mine. The tools just let me finish what I started without needing permission from a system designed to say no.

That is not theft. That is a superpower. And it is coming for everyone who has a story they have been carrying and no way to tell it. The path to an audience is wider than it used to be. That is not a threat to anyone who creates real value. It is only a threat to friction that adds cost without adding quality.

Why I'm Sharing Every Cost: Driving Down the Budget While Raising the Bar

You can see the production cost breakdown under Behind the Scenes on this site. Every dollar. Every line item. Actual spend and estimates for everything ahead. I'm sharing this for two reasons.

First: I want to drive costs down with every film I make while increasing quality at the same time. The only way to do that is to measure honestly. If I know the teaser trailer cost $400, I can ask: what made it cost that? What took the most iterations? Where did I waste generations? Next time, with better prompts, tighter references, and more experience with each tool, that number drops. Same quality or better, less money. That's the cycle. Transparency forces discipline.

The tools themselves are getting better too. What costs $6,000 in compute and iteration today will cost less in six months. Resolution goes up. Consistency improves. Generation speed increases. By publishing real numbers now, I create a benchmark. Future projects can be measured against it. That's useful to me and to anyone else trying to do this.

By late July, when full production starts, I expect the price-to-quality ratio to shift again in my favor. The models improve monthly. Waiting to shoot until the tools are better isn't laziness - it's strategy. The teaser trailer proves the concept now. The full film benefits from every improvement between now and then.

Second: I'm likely taking Young Cop to Kickstarter after the teaser trailer drops on July 4. The trailer is the proof of concept. Once people can see what this looks like, hear how it sounds, and feel the tone, I want to give them a way to be part of finishing it.

I have a defined delivery timeline. I have a high-level plan for tiers and goals. The budget is transparent. The roadmap is public. Backers will know exactly what their money funds and when each piece delivers.

In return, much of the project will remain Creative Commons for personal use. The novel stays free. The soundtrack stays free. Backers get early access, behind-the-scenes content, credits, and the satisfaction of helping prove that one person with the right tools can build something real.

This isn't about getting rich off a Kickstarter. It's about covering the AI compute, voice generation, and cloud rendering costs that turn a story into a finished film. The creative work is mine. The labor is the tools. The funding covers the tools.

More details after the trailer. For now: the numbers are real, the timeline is real, and the work is happening every night.

One Story, Every Form It Can Carry: The Verticalization Strategy

The idea behind Young Cop started small. A character. A situation. A tone. The kind of thing that could have been a short story and stopped there. But it didn't stop. It grew vertically.

Verticalization means taking a story idea and stretching it through every form it can sustain. Short story becomes novel. Novel becomes screenplay. Screenplay becomes film. Film becomes audiobook. Each step isn't a remake. It's a new expression of the same core.

Not every story can do this. Some ideas are a song and nothing more. Some are a film that would make a terrible book. The question at every level is: does the material earn the next form? Does it have enough depth, enough emotional range, enough world to justify another telling?

Young Cop does. A boy in a cruiser at 2 AM, carrying weight nobody asked him to carry. That core translates. It works in prose because you can live inside his head. It works on screen because the visual loneliness is cinematic. It works as music because the grief and faith and silence all want to be sung.

The goal is to take it as far as it feels like it has a place to find an audience. Maybe the novel connects and nothing else does. Maybe the film is the version that breaks through. You don't know until you build each one and put it in front of people.

This is the model. One story idea, explored vertically through every channel that fits. The audience finds it wherever they are. The story meets them there.

One Person, No Crew: What AI Filmmaking Actually Looks Like Day-to-Day

I work in IT full time. Every frame of this film is made between 10 PM and 2 AM. There is no production schedule. There is no crew. There is whatever I can carve out after the workday ends and everyone else is asleep.

Before this job, I worked in Hollywood at a high level for nearly a decade - attending industry events, interacting with talent and executives across film and television, above and below the line. I know what professional production looks like from the inside. This is not that. This is one person, a laptop, and a story that won't let go.

A typical night: I spend two hours writing and refining voice direction for three scenes of dialogue. Then an hour generating footage in Higgsfield, reviewing takes, regenerating the ones that don't match the tone. Then an hour placing sound effects in Resolve, adjusting timing by frames, layering ambient audio until the scene breathes.

Then I hit a consistency problem. A character's shirt changes color between shots. I regenerate. The new take has better consistency but worse framing. I regenerate again. Third time works. That's 45 minutes for one usable clip.

The decisions don't automate. Does this scene need a closer shot? Is the pacing too slow here? Should Marcus speak before or after the silence? Is this the right track from the soundtrack for this moment? Those are all human calls, made dozens of times a night.

The AI handles labor. Rendering frames. Synthesizing voice. Generating orchestral arrangements. The creative direction is constant, manual, and exhausting in the same way any filmmaking is exhausting. You're making a movie. You're just making it with different tools, in the hours most people are asleep.

Average production session: 4-5 hours after work, longer on weekends. About 3-5 minutes of finished, polished film output per day. That's the real ratio.

Why Should You Care? The Emotional Promise of Young Cop.

You are tired. You have 90 minutes. You could watch anything. Why this?

Here is what I can promise you: this film is about a boy who carries a weight no child should carry, and he does it anyway because no one else will. It is about the three people who see him doing it and decide, one by one, to stop letting him do it alone. It is about what happens to a town when one person refuses to look away.

It is not fast. It is not loud. It does not have explosions or chase scenes or a twist you did not see coming. What it has is the feeling of watching someone be brave when they are terrified, and the slow accumulation of small kindnesses that eventually become a family.

If you have ever felt alone in something that was too big for you. If you have ever stayed up past midnight doing work no one asked you to do because someone had to. If you have ever been the person who noticed what everyone else agreed not to see. This film is about you. That is the promise.

The novel tells this story in prose. It is free. Start there if you want. But the film is the version that will sit in your chest. Because silence on a page is just white space. Silence on a screen, with a boy's breathing and the sound of rain on tin, is something else entirely.

The Soundtrack: What Each Track Scores and Why It Matters

Eight tracks. Each one written on piano, fed into Suno, and directed into a full orchestral arrangement. Here is what they are and where they live in the film:

But first: the piano. Every track started on my piano. My hands. My melody. That recording - the raw piano take - is the seed that everything else grows from. And I made a deliberate choice: the piano stays audible in the final arrangement. It is not buried under the orchestra. It sits inside the strings and brass the way a heartbeat sits inside a chest. You can hear the human hand that started it. The imperfections. The breath between phrases. The orchestra surrounds it but never replaces it. The piano is Caleb. The orchestra is the town. One lives inside the other.

01 - Ten and Two. Dm, 3/4, 149bpm. The opening patrol. Caleb's hands on the wheel. Quiet strings rising slowly. The theme that says: this is what it sounds like to be awake when the world isn't.

Full chord chart for Ten and Two (c) Keith Adler:The letters (A, B, C, D) are song sections - like chapters in a story. B is the verse (where the melody lives), A is the chorus (where it opens up), C is the pre-chorus (the breath before), and D is the bridge (the departure that makes coming home mean something).B m.1Dm | Dm | Gm/D | Dm Gm | Gm | Gm | Bb/F | Em7 | A7 | A7 | Dm Dm | Gm | C7 C | Fma7 | Bbma7 | Em7 | Asus4 | A | Dm | GmA m.21C | Fma7 | Bbma7 | Edim | Asus4 A7 | Dm | Dm | Gm | Gm7 | C | C | F | F | Bbma7 | Bbma7 | Gm | Gm | Em7 | Em7 | Asus4B m.41Asus4 | A | A | A7 | Dm | Dm | Gm7A m.48Gm7 | C | C | Fma7 | Fma7 | Bb7D m.54Bb | Bb | Gm7 | Gm7 | Em7 | Em7 | Asus4B m.61Asus4 | Asus4 | Dm | Dm | GmA m.66Gm | C | C | F | F | Bbma7 | Bbma7 | Gm | GmB m.75Asus4 | Asus4 | Asus4 | Fma7 | Dm Bb7 Dm | Dm | Dsus4 | Bb | Bb/D | C7 A7 | A7 | Dm | Dm C6A m.88Bb | Bb | Csus4 | C | F | F | F#dim | AmB m.96A | Asus4 | Dm Bb/D | C7 | Fma7 | Em7 Bbma7 | Bb | A7 | A7 | A | A | F/A | E | D | Dm | Gm | A7 | A7 | A7 D | Dm | Dm | Dm | Dm | N.C.

The harmonic language tells you everything. The piece lives in D minor but keeps reaching toward major - Fma7, Bbma7, C - like a boy who wants warmth but keeps ending up back in the dark. The suspended chords (Asus4 everywhere) never resolve cleanly. They hang. The way a question hangs in the air when no one answers the radio. The bridge drops to Bb and Gm7 - the furthest harmonic distance from home - which is where the danger lives. Then it pulls back through Em7 to that Asus4 again. Always suspended. Always unresolved. Until the very end: N.C. No chord. Silence. The patrol is over. The engine cuts. The music stops the way a shift ends - not with resolution, but with absence.

02 - The House That Waits. F major, 76bpm. Caleb's empty house. Cold soup. The answering machine. Piano and nothing else. The loneliest track on the record.

Full chord chart for The House That Waits (c) Keith Adler:B m.1D | Dm | Dm | Fma7 Fma7 | Bb | Bb | F | F | Gm | Gm | DmA m.12Dm | C | C | C | Dm7 Dm | Dm | C6 | Bbma7 Bb | F | C/E | Dm7 Dm | Dm | F/A | Am | BbC m.27Bb | Gm | C7 C | Fma7 | C/E | Dm7 Dm | Gm | C | F | F | Bbma7 | C | Am7 | Dm7 | Bb | C | Am C | DmA m.45F Fma7 | Am | Bbma7 Bb | Bb | C7 | Dm | C6 | Bb | Bbma7 C | Dm7 | C6 | Bbma7 | F | C/E | Dm | C6 | Bbma7 Bb | Bbsus4 C | Dsus4 Dm | Bbma7 | Bbma7

The House That Waits is listed as F major but listen to where it actually lives: D minor. The very first chord is D major - bright for one beat - then immediately Dm. The brightness was a memory. It is gone before you can hold it. The B section walks through Fma7 and Bb (the relative major territory) but always returns to Dm. The A section is the most harmonically active - Dm, C, C6, Bbma7, F, C/E - the piano walking through the house the way Caleb walks through it. Room to room. Not looking for anything. Just moving because standing still is worse. The C section is the longest - it is the middle of the night when sleep will not come. Bb to Gm to C7 to Fma7 - constantly shifting, never landing. The final A section ends on Bbma7 repeated. Not Dm. Not home. The piece does not resolve to its own key. It just stops in a place that is almost warm. Almost. That is what the house is. Almost a home. Not yet.

03 - Third Friday. F major, 77bpm. The stakeout. Four hours of watching. Low tension building. The track that plays under silence and makes the silence worse.

Full chord chart for Third Friday (c) Keith Adler:C m.1Dm | Dm | Fma7 | Bbma7 Bb | Bb | F | C/E | Dm7 Dm | AmB m.10Am Bb | Bb | C7 | C | Dm | C6 Am | Bbma7 | Bbma7 | Fma7 Fma7 C | Dm | Am | Bbma7 Bb | Csus4 | Csus4 C | Dm | Am | Bb | Bbma7 F | F/C C | Dm Dm | Am | Bbma7 BbA m.33C | Fma7 F | C | Dm Dm | Am | Bbma7 Bb | F | Gm | C | F | F/C C | Dm7 DmB m.45Am | Bb | Csus4/C F C6 C | Dm | C6 | Bb | F | Csus4 C | Dm | Am | Bbma7 Bb | Csus4 CC m.57C F | F7 | C | Dm | Dm Am | Am BbB m.63Bb C | F/C | Dm | Dm C6 | Bbma7 | Csus4 Csus4 F | F | N.C.

Third Friday is almost the same tempo as The House That Waits (77bpm vs 76bpm) and that is deliberate. Waiting alone in a car watching a door is the same loneliness as being alone in a kitchen at 2 AM. The body does not know the difference. The C section opens with Dm reaching immediately toward Fma7 and Bbma7 - the major chords that live in F major's orbit. Hope. Routine. The town is quiet. The B sections are where the tension lives - Am to Bb to C7, circling without resolution, the way four hours of watching nothing circles. Csus4 appears constantly - suspended, never landing. The A section is the one moment of forward motion in the whole piece: C to F to Gm to C to F. A sequence that actually moves somewhere. This is when the radio crackles. When something finally happens. Then the final B section lands on F - not Dm. The piece ends in the major key. Because the stakeout ends with the call. The waiting is over. Something broke. N.C. Go.

04 - Six Blocks East. Am, 95bpm. The break-in at the hardware store. Caleb driving fast for the first time. Urgency without panic. The case is real now.

Full chord chart for Six Blocks East (c) Keith Adler:B m.1Dma7 | Bbma7 Am | Am | D | Am7 | Am | B7 | A | Am Am | Am | AmA m.12Am | Am | Asus4 A | Ab | Abm | Ab | Abma7 Am | Am | Asus4 AC m.21F | F | Am | AmA m.25Abm | Abdim | B6 Am | Am | Abm | Abm | Am | AmC m.33F | F | Dm D | Am | Am | F | F | DmB m.41Am | Am | Dm7 | Db7 | Db7 | Bb | Am | AmA m.49Ab | Ab | Am | Am | Ab | Ab | AmC m.56Am | F | Dm | Am | Am | F | DmB m.63Am | Am | F | Fm | D | D Dma7 | A | A | N.C. | N.C.

Six Blocks East is the first time the soundtrack leaves the world of F major and D minor. A minor is the relative minor of C major - a different harmonic universe from the first three tracks. The case has broken open. We are in new territory. The piece is defined by Ab - a chord that has no business being in A minor. It appears in the A sections like a shadow. Abm, Abdim, Abma7 - chromatic visitors that make the harmony feel unstable and dangerous. Something foreign has entered Princeton. The B sections keep returning to Am with Dm7 and Db7 - flat-seven chords that pull the floor out. The C sections are simple (Am - F - Dm) like Caleb trying to think clearly while everything accelerates. The final B section ends: F - Fm - D - D Dma7 - A - A - N.C. The shift from F to Fm (major to minor on the same root) is the moment of recognition. Then D major - bright, sudden, certain. Then A major. Not Am. Major. He knows. N.C. He is already driving.

05 - Nice Car. E major, 98bpm. The Polaroid. The threat. Something dark in the brass. The track that says: they know where you park.

Full chord chart for Nice Car (c) Keith Adler:B m.1N.C. | C E | A/E F | E | A# | A# | A/E E | E | C/E C | C/E | E/B | E | EC m.14E | Esus4/E | C/E | Gm Esus4 | C/E | Gm Em | EmD m.21E | E | D#7 E | E | D#m E | E | F#7 | D# C/E | E E | A#7 CmB m.31E | Em | E | E | G7 | E | Em | Em | E | Esus4/F# D Ama7 A# BG | Esus4 | E7 | E7 | Em | Em | C7D m.47E | E | D# E7 | Bb E | D#m E | Em | Em | Cma7 | Cma7 C7A m.56Am | Am | B | B | B | BB m.62G E7 | E | F# | E | E | F# | E | F#7 | G D# | D# | E | E | E | E7 | N.C. | N.C.

Nice Car is the strangest piece on the record. E major at 98bpm - the only track in a sharp key. Everything else lives in flats (Bb, Eb, Ab). E major is bright. Confident. That is Reno. He is not afraid. The threat is delivered with a smile. The piece opens N.C. - no chord - then drops into C E with chromatic movement (A#, D#, F#) that has no harmonic logic except menace. The C section introduces Gm inside an E major piece - a chord from a completely different universe. Something is wrong. You can feel it even if you cannot name it. The D sections use D#7 and D#m - the leading tone of E - circling the home key without ever quite being in it. Like someone standing just outside your house. The A section is shockingly simple: Am - Am - B - B - B - B. The mask drops. Minor to major dominant. No decoration. Just pressure. The final B section ends with chromatic descent through F#, G, D#, then bare E chords. Then E7. Then N.C. The dominant seventh hanging unresolved into silence. The threat does not end. It just stops being audible.

06 - Dragons. Am, 100bpm. Ricky's drawing. The classroom. The moment of connection before the loss. Warm and brief. A melody that doesn't finish. The piece I'm proudest of on this record.

Full chord chart for Dragons (c) Keith Adler:B m.1Am | Am | Dm | Dm | A | A | A | A AmC m.9Am | Am | Am | Am | Am | AmA m.15F | F | Fma7 G6 | G | GC m.20Am | Am | Esus4 Am | Am | Dm | DmA m.26G | G | Em E | E | Am | Am | G F | F | G | GD m.36E7 | E7 | Am | Am | Dm | DmA m.42G | G | Cma7 | Cma7 | Fma7 | Fma7D m.48Dsus4 Dm | Dm | E | E | EB m.53Am | Am | Dm A | A7 A7 | A | A

Dragons is harmonically simpler than Ten and Two - and that is the point. The C sections are just Am repeating. Six bars of the same chord. A boy passing a drawing to another boy. Nothing complicated. Just presence. Then the A sections reach outward - F, Fma7, G6 - warmth arriving without announcement. The way kindness does when you are not expecting it. The E7 in the D section is the turn - dominant of Am, pulling you back into the minor inevitability. You know what is coming for Ricky. The harmony knows too. It reaches for major (Cma7, Fma7) and then Dsus4 pulls it back down. The final B section ends on bare A major chords - not Am. Major. Because in Caleb's memory, this moment stays warm even though everything after it went dark. The melody does not resolve because the story did not let it.

07 - The Hollow Bell. Em, 134bpm. The raid. The silence. The aftermath. The biggest piece on the record. Builds to nothing and stays there.

Full chord chart for The Hollow Bell (c) Keith Adler:B m.1F/G# | F | Em | Em | Em | C | CA m.8C | D | D | D | G | G | D/F# | D/F# | C | C | C | Am | AmC m.21Am7 | Bsus4 | B | Bsus4 | Bsus4 | B | BB m.28B | Em Em | Em | Am | Am | D7 | D | DA m.36G | Db B | Cma7 C | C | Am AmC m.41Am | Bsus4 | Bsus4 | B | B | B7 | Bsus4 | Em | EmA m.50Am | Am | D | D | G | G | C | CC m.58Am | Am | Bsus4 | Bsus4 | B | B | Am | Am | D7 | D | DG | G | Cma7 | C | C | D/F# | Am F#dim | F#m BB m.77Bsus4 | Em | Em | Em | EmD m.82Am | Am | D | DA m.86Gma7 | G | G | Cma7 | Cma7 | Am | AmB m.93B7 | B | Bsus4 | Cma7 | Cma7 | Bsus4 | F | EmD m.101Am | Am | D | DA m.105Gma7 | G | G | Cma7 | C | CC m.111D/F# | D/F# | B | BB m.115E7 | Em | Em | Em | Em | N.C.

The Hollow Bell is the longest and most structurally complex piece on the record. It is also the darkest. E minor at 134bpm - the fastest tempo on the soundtrack except Stay Gold. It opens with F/G# - a chord that does not belong in E minor. It is wrong on purpose. The raid has not started yet but something is already dislocated. The A sections climb through major territory (G, D, C) like hope building, but the C sections keep pulling back through Bsus4 - suspended, unresolved, the same trick as Ten and Two but angrier. The B sections anchor in Em with a weight the other tracks do not have. The D sections (Am - Am - D - D) are the simplest harmony in the piece - four chords, repeated - because this is where Caleb is just breathing. Just existing in the aftermath. The final B section: E7 - Em - Em - Em - Em - N.C. The dominant seventh resolves to minor and then stays there. Four bars of Em. Then nothing. This is the sound drop. The moment all sound disappears from the film. The music does not fade. It stops. Like a heartbeat that was there and then was not.

08 - Stay Gold. Dm, 137bpm. The morning after. French toast. Sunlight. Coleman's table. The theme from track one, resolved. Finally warm.

Full chord chart for Stay Gold (c) Keith Adler:C m.1Dm | Dm | Gm | GmB m.5C | Cma7 C7 | F | Fma7 Fma7 | Fma7 | Bbma7 | Bbma7A m.12Gm | Gm | Asus4 | Asus4 | Asus4 | A | AC m.19Dm | Dm | Gm | GmB m.23C7 | C | C | Fma7 | F | F | Bbma7 | Bb | Gm7A m.32Gm7 | Asus4 | Asus4 | A | AC m.37Dma7 | Dm | Dm | Dm Gm | GmB m.42C | C | F | F | Bb | Bb | Bb7 Gm | GmA m.50Asus4 | Asus4 | Asus4 A | AC m.54Dm | Dm | Gm | GmB m.58C7 C | C | Fma7 | F | F | Bb | Bb | EA m.66Gm | Gm7 | Gm Asus4 | Asus4 | A | AB m.72A7 Gm7 | Gm7 | Csus4 | C | C | Fma7 | Fma7 | Bbma7 | Bb | Bb | Gm | GmA m.84A7 | Asus4 | AD m.87Dm | A/F | D/F# | D/F# | Gm | Gm | C7 | C | CB m.96F | F | Bbma7 | Bbma7 | Bbma7 | Ebma7 | Asus4 | Asus4C m.104A | A | Dm Dm | Dm | Gm | C | Fma7 | Bbma7 | C/E | Gm7 | Asus4 | AB m.116D | Dm Dm | Dm | Dm | Dm | Dm | N.C.

Stay Gold is the resolution. Same key as Ten and Two - D minor - but faster (137bpm vs 149bpm in 3/4, which makes Stay Gold feel more driven in 4/4). It opens with the C section: just Dm and Gm. The same two chords that started the whole film. But now they are not lonely. They are a doorstep. The B sections climb through major territory - C, Fma7, Bbma7 - with confidence the opening track never had. The A sections keep landing on Asus4 resolving to A - the dominant of Dm - but here the resolution feels earned instead of suspended. The bridge (D) introduces A/F and D/F# - chromatic bass movement that lifts the whole piece off the ground. It is the one section in the entire soundtrack that feels like flying. Then the final C section does something no other track does: it walks through Fma7 - Bbma7 - C/E - Gm7 - Asus4 - A. Every major chord in the key, in sequence, like a hand touching every person who mattered. Then D minor. Then silence. The same ending as Ten and Two. N.C. But this time the silence is not absence. It is peace.

The novel has no soundtrack. The film does. Music is the layer that tells you how to feel about what you are seeing. These eight tracks are that layer. They are free to stream on the site.

A note on the keys and tempos. This is deliberate. The film opens in D minor at 149bpm - that's a waltz tempo in a minor key. It moves. It's restless. Caleb is awake and the town isn't. The car is rolling. The music needs forward motion with unease underneath.

Then the house: F major, 76bpm. Half the speed. A major key that should feel warm but doesn't because it's so slow and so alone. The emptiness is in the tempo, not the harmony.

The stakeout stays in F major but at 77bpm - almost identical to the house. Because waiting alone in a car and being alone in a kitchen are the same loneliness. The music knows that even if Caleb doesn't.

When the case breaks open - Six Blocks East - we shift to A minor at 95bpm. Faster. Darker. The relative minor of the stakeout. Same harmonic family, different weight. Something has changed.

Nice Car pushes to E major at 98bpm. The threat. Major key because Reno is confident. He's not scared. The brightness is menacing. That's the trick - a major key at the wrong moment feels worse than minor.

Dragons - A minor, 100bpm. The connection with Ricky. Back to minor because this moment of warmth exists inside tragedy. The melody doesn't resolve. It can't. We already know what's coming.

The Hollow Bell blows open: E minor, 134bpm. The raid. The fastest and darkest point. Everything the film has been building toward. Then it stops. The silence in the film happens here. The music builds to the moment it disappears.

And Stay Gold brings it home: D minor, 137bpm. The same key as the opening. Full circle. But faster now - because forty years have passed and the boy became the man and the man still drives the same road. The resolution is not rest. It's continuation. He's still checking. The tempo says so.

Content Advisory: What's in the Film and What Isn't

For parents, educators, and anyone who wants to know before they watch:

What is in the film: A 14-year-old carries a gun (he never fires it). Drug dealing is depicted (no use shown on screen). A woman overdoses (off-screen, reported via police scanner). A man threatens a child verbally. An arrest scene with shouting and physical restraint. Mild profanity from adult characters. A child eats alone, lives alone, and is in emotional distress throughout.

What is NOT in the film: No graphic violence. No gunfire. No blood. No sexual content. No nudity. No slurs. No torture. No harm to animals. No suicide or self-harm.

Thematic content: Abandonment. Grief (implied, not depicted). Institutional failure toward a child. Drug addiction shown through physical deterioration. A community choosing not to see what is happening. Adults failing to protect a child they claim to love.

Recommended age: 13 and up. Younger viewers who are strong readers may engage with the free novel first, which covers the same events in less intense form. The film's emotional weight comes from sound, silence, and duration - things that hit differently than words on a page.

This is not a violent film. It is a heavy one. The weight comes from caring about a boy who is carrying too much. That is a feeling, not a content warning. But I want you to know what you are walking into.

Character Consistency Across 150 Images: Maintaining Visual Identity When Every Frame Is Generated Independently

The hardest problem in AI filmmaking isn't generation quality. It's consistency. Every image is generated independently. The model doesn't remember what Caleb looked like in the last frame. You have to make it remember.

The approach: locked reference images for every character. Caleb has a primary reference sheet. Front, side, detail of the badge, detail of the uniform. Every generation prompt includes these references. Every output gets checked against them.

When something drifts, it usually drifts subtly. Hair slightly different. Nose shape shifts. Jacket color moves a shade. You catch these by looking at sequences of images side by side. If you can't stack three frames and have the character read as the same person, it goes back.

Locations are easier. A diner doesn't change expression. But they still drift in architectural details. The number of booths. The position of the neon sign. The color of the counter. Reference sheets for locations too.

Props are the easiest because they're simple objects. Badge. Notepad. Phone book. Cruiser. But they matter the most for continuity. If the badge is on the wrong side in one shot, every viewer notices.

The production design library is 150+ images. All of them locked as references before film generation began. That's the foundation. Without it, you don't have a film. You have a collection of unrelated pictures.

Working with Suno: Directing Music Generation Like a Composer Directs an Orchestra

Suno doesn't give you what you want on the first try. It gives you something in the neighborhood. Your job is to know which neighborhood you're looking for and keep iterating until you're standing in front of the right house.

Every piece was played by me first. Keyboard. I had very specific keys and BPMs in mind for each track before I ever opened Suno. The compositions existed. Suno brought them to where I envisioned them: full orchestral arrangements I couldn't produce alone.

My process: I compose the structure on keyboard first. Not a full arrangement. A melody, a chord progression, a rhythm. Then I describe what I want in precise musical language. Not "make it sad." That gives you generic. "Warm strings in the mid-register, piano leading with a descending figure in D minor, tempo 72, no percussion for the first 16 bars." That gets you somewhere specific.

Each track took between 15 and 40 iterations. Some came fast. "Ten and Two" landed on the sixth try. "Stay Gold" took almost 40. The difference is how specific your ear is about what's right versus what's close.

I treated Suno like a session musician who's incredibly talented but needs clear direction. Vague direction gives you vague results. Specific direction, musical terminology, reference to tone and texture rather than mood words: that's how you get a score that sounds intentional.

The 1980s orchestral style helped constrain the space. Once you know you want Spielberg-era warmth, you can reject anything that sounds modern, synthetic, or overproduced. The constraint is a creative tool. It makes the decisions faster.

Nine tracks total: eight orchestral pieces plus the duet. All of them sound like they belong to the same film. That's not luck. That's iteration and refusal to accept "close enough."

The Audiobook as a Separate Product: Same Story, Different Experience

The audiobook isn't an afterthought. It's a separate telling of the same story, designed for a different mode of attention.

When you read the novel, you control the pace. You can pause, reread, sit with a sentence. When you listen to the audiobook, the narrator controls the pace. That changes everything about how the story hits you.

Caleb's voice narrates. The same ElevenLabs voice profile used in the film. So when someone listens to the audiobook and then watches the movie, Caleb sounds like the same kid. That continuity matters.

The narration performance is different from film dialogue. Film dialogue is terse. Audiobook narration breathes. It takes its time. It lets silence exist between paragraphs. The internal monologue that the film can't carry lives fully in the audiobook.

The audience for audiobooks is different too. Commuters. People who listen while working. People who don't have time to sit and read but have hours of audio time in their day. That's a real audience that novels and films don't reach.

Same story. Same world. Different door in. That's the point of multi-format storytelling. Meet people where they already are.

Why I Built a Production Manager Instead of Using Spreadsheets

Young Cop has 79 scenes, 150+ production design assets, 16 release milestones, 8 soundtrack tracks, character voice profiles, location databases, and prop inventories. A spreadsheet breaks at this scale. Not technically. Mentally.

So I built a custom production manager. A web app. It lives at the same URL as the public site, behind a login. It tracks every scene: status, characters involved, location, time of day, props needed, dialogue status, voice generation status, footage status.

I can see at a glance which scenes are fully produced, which are waiting on voice work, which need regeneration. The cast page pulls from the same data. The stills gallery is managed from there. Everything is one source of truth.

Built it with Claude. Node.js backend, single HTML frontend, JSON database. No framework overhead. It does exactly what I need and nothing I don't.

The reason this matters: when you're one person doing everything, you need a system that shows you what to do next. Not what you've already done. Not the whole picture. Just the next action. The production manager does that.

Could I have used Notion or Airtable? Sure. But then I don't own it. I can't customize it. And when I need a feature that doesn't exist, I can't build it in 20 minutes. This way I can.

Writing Dialogue That Sounds Like 1985: No Cell Phones, No Internet, No Shortcuts

The hardest part of writing period dialogue isn't avoiding anachronisms. It's understanding how the absence of technology changes how people talk to each other.

In 1985, if you needed to reach someone, you called their house. If they weren't home, you drove there. If you had information to share, you said it face to face or you waited. That changes the rhythm of every conversation.

People in 1985 explained things to each other. They gave directions. They described what they saw because they couldn't send a photo. They made plans in advance because they couldn't text "running late." When someone didn't show up, you worried. Actually worried.

Caleb's dialogue reflects this. He reports what he sees on patrol in full sentences over the radio. He leaves notes. He tells people things in person that a modern kid would text. That physical presence in every interaction gives the dialogue weight.

The other thing: regional speech patterns. Small-town Kentucky in 1985. People don't talk like a screenplay. They trail off. They start sentences they don't finish. They reference shared context without explaining it because everyone in town knows everyone.

I read transcripts of real Kentucky speech patterns from the era. Court transcripts, oral histories, local radio archives. The goal wasn't phonetic dialect writing. It was rhythm. How people breathe between words. Where the pauses go. What gets left unsaid.

When Marcus says "You're not fine. But you're there." - that's 1985. Direct. No emoji. No subtweet. Just a man telling a kid the truth because there's no other way to do it.

The Real Risks of Making an AI Feature Film - And How I'm Addressing Them

I've seen the skepticism. It's fair. Let me address the biggest risks people have raised about this approach, because I think about them constantly.

Character Consistency. This is the number one killer for long-form AI film. Even with locked references, Caleb's face, proportions, and the fit of that oversized uniform will drift across scenes. I know this. My approach: every character has a primary reference sheet - front, side, details of key features. Every generation prompt includes these references. Every output gets checked against them. I do frame stack comparisons across scenes to catch drift before it compounds. It's going to be painful at minute 40. I accept that. The answer isn't hoping it won't happen - it's building a review pipeline that catches it immediately and regenerates before moving forward.

Tone and Atmosphere Drift. Southern noir lives on mood. Rainy streets, quiet dread, that specific 1985 Kentucky lighting. Maintaining the same visual tone across hundreds of generations is extremely hard. My fix: I'm building scene-by-scene tone references. Not just character references - environmental references. What the rain looks like. What the streetlight color temperature is. What the cruiser dashboard glow does at night. Every generation gets checked against the tone sheet for that location and time of day.

Time and Burnout. I'm doing this solo while working full-time. Generating, reviewing, rejecting, and regenerating over a thousand clips at night is a brutal pace. The teaser trailer is going to be the first real stress test. My answer: I don't rush. The timeline is aggressive but not suicidal. If something takes longer, it takes longer. I'd rather ship late than ship bad. And the workflow itself is designed to reduce decision fatigue - the script is locked, the production design is locked, the references are locked. I'm not making creative decisions at generation time. I'm executing against a plan.

Editing Load. Even if the footage looks good, stitching it all together into something that feels like one cohesive movie in DaVinci Resolve is where a lot of solo AI projects fall apart. This is why I spent so much time on the script structure. The edit is not improvisational. Every scene has a defined in and out. Every transition is planned. The sound design carries continuity across cuts. When two shots don't match perfectly, the audio layer - ambient sound, score, foley - creates the through-line that your brain follows.

Here's my honest assessment: feature-length AI film is still uncharted territory. Shorts are proven. Features are pushing the edge. I know that. But the story is strong, the infrastructure is built, and the workflow is designed for exactly this kind of sustained effort. The question isn't whether there will be pain. There will be. The question is whether the pain produces something that works. I think it will.

The consistency battle is real. I'm not pretending otherwise. But I'd rather fight that battle with a locked script, locked designs, locked references, and a clear pipeline than wing it like most people do. That's the difference between a project that finishes and one that doesn't.

The CC BY-NC 4.0 Decision: Why I'm Giving the Novel and Soundtrack Away for Free

The novel is free. The soundtrack is free. Both under Creative Commons Attribution-NonCommercial 4.0. Anyone can download them, share them, remix them for non-commercial use. Here's why.

The traditional model: you make something, you charge for it, you hope enough people buy it to break even. For an unknown indie filmmaker with no audience, that model means nobody sees it. The friction of a paywall kills discovery.

The alternative: give away the things that build audience. The novel hooks readers. The soundtrack hooks listeners. Both point back to the film. The film is where the long-term value lives: adaptation rights, commercial licensing, stage rights.

CC BY-NC means people can share it freely, but they can't monetize it. No one puts my soundtrack in their YouTube video for ad revenue. No one sells prints of the concept art. Non-commercial means the audience grows but the commercial rights stay with me.

There's also a practical angle: if someone infringes, CC BY-NC gives me clear legal standing. The license terms are unambiguous. The boundary between free use and commercial exploitation is bright and specific.

The screenplay and the film stay All Rights Reserved. Those are the crown jewels. The novel and soundtrack are the front door. You give away the front door. You own the house.

A free novel that 10,000 people read is worth more than a $5 novel that 200 people buy. The 10,000 readers become an audience. The 200 buyers stay strangers.

Building the Entire Website and Production Pipeline with Claude

Everything you're looking at right now was built with Claude. The website. The production manager. The CDN pipeline. The deployment system. The API. All of it.

I'm not a frontend developer. I'm a creative. Tech is my superpower for conveying stories, whether films or songs. Claude fills the gap between what I can describe and what I can build. I say what I want. Claude writes it. I review, adjust, ship.

The site is a Node.js Express app on Fly.io. Single HTML file frontend. No React, no framework, no build step. Assets on Cloudflare R2. CDN caching. The whole thing deploys in under 30 seconds.

The production manager behind the login is the same app. Scene tracking, character management, image uploads, PDF generation, data exports. All custom, all built in an ongoing conversation with Claude. It's a living process - I add features as I need them, sometimes daily.

This matters because the alternative was hiring a developer. Or spending months learning a framework. Or using a template that doesn't do what I need. Instead: describe the feature, get the code, test it, deploy it. Same day.

The meta lesson: the same AI-assisted workflow that makes the film possible also makes the distribution infrastructure possible. One person can build all of it now. The tools exist. The barrier isn't technical anymore. It's creative.

Total cost for all website development: under $100. Just the Claude subscription I was already paying for the writing and production work.

The Goal: One Story, Every Channel It Can Reach

Here's what I'm trying to do. I have a story. I want to take that story and put it into every form that makes sense: novel, soundtrack, film, audiobook. Not because every story deserves all of those. But because when a story is strong enough, it should have every possible path to find its audience.

I'm not trying to type a movie idea into a machine and have it come out the other side. That's not what this is. I want to be the pilot of a modern vehicle. The story is in my head. The AI tools are the instruments that let me take what's in my head and share it with the world at a scale one person has never been able to do alone.

I have multiple story ideas. Young Cop is the one I think I can stretch the most across channels. The characters are deep enough. The world is specific enough. The emotional core translates across mediums. A 14-year-old cop in a dying Kentucky town works as a novel. It works as a film. It works as an audiobook. Each form finds a different audience.

Maybe it falls flat wholesale. Maybe it only finds a place in book form. Maybe the film connects and nothing else does. I don't know. But the story sees the light. It has a path to an audience. That's what matters.

I want to invest fully in the story and the world. Not spread thin across half-finished projects, but go deep on one thing and see how far it can go. AI lets me do that without a studio, without a team, without permission. The creative energy is mine. The tools handle the labor. The audience decides what sticks.

Ethics and AI: An Enabler for Sharing a Story, Not a Replacement for Making One

I want to be clear about how I use AI on this project. It is not driving the story. It is not writing the characters. It is not making creative decisions. AI is an enabler. It lets me share a story that already exists in my head with an audience that would never see it otherwise. The vision is mine. The tools are labor.

More than that, I use AI adversarially. When the novel was finished, I ran it through multiple AI systems and told them to tear it apart. Find the contradictions. Find the dropped threads. Find the moments where a character acts against their established motivation without justification. I wasn't asking AI to write. I was asking it to challenge. To stress-test. To make sure the story I built can survive scrutiny.

That adversarial approach extends to every part of the process. When I generate a scene, I'm not accepting what comes out. I'm directing, rejecting, iterating, and choosing. The AI proposes. I decide. Dozens of times per scene. The creative authority never leaves my hands.

There's another ethical dimension that people don't talk about enough: what AI visualization removes from reality. Every location in this film is generated. That means no real storefronts. No real brand logos. No trademarked signage. No identifiable businesses. The AI creates spaces that feel like a convenience store in 1985 Kentucky without depicting any actual store. It creates an outfit Caleb would wear without copying a real brand's design.

This is a feature, not a limitation. Traditional film production spends significant time and money clearing logos, covering brand names, negotiating usage rights, or digitally removing trademarks in post. AI-generated imagery bypasses all of that by creating from scratch. The diner has a neon sign, but it's not a real diner's sign. Coleman's store has shelves of products, but none of them are real products. The world feels authentic without infringing on anyone's intellectual property.

The same applies to characters. No real person's likeness is used. No actor's face is referenced. Every character is designed from description, not from reality. That's a cleaner ethical position than deepfakes, face-swaps, or unlicensed likenesses that plague other AI projects.

The line I draw: AI handles production labor. The story, the characters, the moral questions, the creative choices, and the accountability all stay with me. If this film succeeds, a human made it. If it fails, a human made it. The tools didn't decide anything. I did.