JORDAN COPELAND

I edit videos. I direct them sometimes. And increasingly I build the generative pipelines that let productions do things their budget says they can't.

The editing is the foundation. Twenty years of it. Short-form, mostly: brand films, music videos, concert films, livestreams, social content, the occasional documentary. Everything from thirty-second Instagram cuts to four-hour live broadcasts.

The directing grew out of the editing. When you spend enough years fixing other people's rushes, you develop strong opinions about what should have been shot differently. Eventually someone lets you prove it.

The generative work is the newest layer. Two years of intensive tooling across sixty-odd platforms, plus a research project called Guava and regular lectures for young filmmakers. The pitch is simple: with the same money, people get to make better things.

The edit is where the work lives. Directing is assembling the conditions for the edit to be good.

When I direct, I'm thinking about what the timeline needs. Which angles give me options, which performances have the energy, whether the coverage allows the cut I can already feel. Most directors think about the shoot. I think about what comes after it.

That's not better or worse. It's just what happens when you've spent two decades inside timelines before you step onto a set.

Generative video tools produce beautiful, unpredictable, frequently wrong output. The industry calls it "slop." I call it raw material.

A Slopwrangler takes generative output and makes it do what the project needs. That means knowing which model produces which kind of thing, how to art-direct a machine that doesn't understand art direction, and when to throw away three hours of generations because the fourth one has something alive in it.

It's editing, really. Just with different inputs.

Incredible: it has to make you feel something. Credible: it has to feel true to whoever's behind it.

A music video for an artist nobody's heard of needs to feel like a discovery, not a marketing exercise. For a global brand, the film has to feel like they believe their own message. And a concert film? It needs to feel like being there, not like watching a recording of being there.

The comma between "incredible" and "credible" is doing a lot of work. They're almost opposites and you need both.

Film school, then a runner job I was terrible at, then six months of pretending I could use Final Cut Pro until I actually could. First real client was the BBC. First real lesson was that nobody cares how clever your edit is if the deadline's passed.

I freelanced for eighteen years out of various East London studios, worked my way through CNN, Nike, Samsung, YouTube. Made two documentaries. Played in several bands at the same time, which is excellent training for the logistics of creative collaboration if nothing else.

A creative editing and video engineering company I ran for four years. Started as a way to formalise the team I'd been assembling informally for a decade.

Tentacle did concert films, fashion films, livestreams, stadium visuals, SaaS product videos, the occasional art project. We pivoted hard into live streaming and remote production during 2020 for obvious reasons, and came out the other side with a speciality in high-stakes live broadcasts.

I closed it to retool for generative work. The company did what it was supposed to do. Four years of managing a team taught me how to brief creative people, how to argue about budgets, and how to ship complicated projects with too many stakeholders. Those skills turned out to be more useful than the company itself.

The list I'm supposed to recite: Adidas, Nike, YouTube, Red Bull, BBC, CNN, Samsung, Channel 4, Universal Music.

The list I actually care about: Alabaster dePlume, GoGo Penguin, The Comet Is Coming, Shabaka, Super Best Friends Club, Primus, Beth Orton, Alewya. Musicians who trust me with their work and let me take risks with it.

Also: two seasons of being embedded with The Police during their reunion tour, which is a documentary story I'll tell you over a drink sometime.

Swedish House Mafia's livestream, probably. Four hours, live to air, five cameras, a warehouse in Hackney dressed to look like a Stockholm basement, and a director (me) making cut decisions in real time while the biggest DJs in the world played to a global audience.

The closest runner-up is three seasons at London Stadium for West Ham. Not technically difficult in the same way, but politically complicated. Every screen in the ground, sixty thousand people, a club with strong opinions, and ninety seconds to land the energy before kick-off.

Summer of 2022. I read the leaked LaMDA transcripts and spent the rest of that week unable to think about anything else. By September I had accounts on every tool I could find. By early 2023 I was using Midjourney to pre-vis a music video. By mid-2024 I was running multi-model pipelines with agentic code managing the prompt strategy.

The acceleration is real. What takes me an afternoon now would have taken a team of three a week in early 2024. What took that team a week was impossible in 2023.

No. This is the wrong question and I'm tired of hearing it.

AI replaces specific tasks that were previously expensive or slow or both. A Slopwrangler doesn't replace a VFX artist. A Slopwrangler does the things a production couldn't afford to hire a VFX artist for. The budget was going to be spent either way. The question is whether it buys twenty minutes of generic stock footage or sixty seconds of something made specifically for this project.

Every technology in the history of filmmaking has triggered this conversation. Editing software didn't replace editors. Digital cameras didn't replace cinematographers. The tools change. The need for someone who knows what to do with them doesn't.

Yes. That's literally it.

A music video with a treatment that requires twelve locations and a budget for four. Generative tools handle the eight locations the production can't physically reach. The director gets the film they wrote. The artist gets visuals that match the song. The budget doesn't move.

A brand campaign that needs character consistency across nine scenes. Previously that's a casting call, a wardrobe department, two shoot days. Now it's one shoot day and a pipeline that extends the results into every scene the budget couldn't cover.

The money doesn't go down. The work gets closer to what it was supposed to be.

Sixty-odd accounts at last count, but the ones that earn their keep:

IMAGE: Recraft V3, Flux, Midjourney v7. VIDEO: Kling, Runway Gen4, Seedream. AUDIO: ElevenLabs, Udio. EDIT: Premiere, Resolve, TouchDesigner. 3D: Blender, Notch, Unreal. LANGUAGE: Claude, GPT, Gemini.

The landscape shifts every few weeks. I'm not loyal to platforms. I'm loyal to results.

An HCI research project. The short version: I'm studying how creative professionals actually interact with generative tools, what works, what fails, and where the friction is.

The longer version involves a Hetzner VPS, several Telegram bots, a personal knowledge system called Wisdom, and more Python than I expected to write in my forties. If you're interested in the methodology, ask me at a lecture or buy me a coffee.

Every atom in the universe is heading towards entropy, which is to say a nice lie down. Making things is the opposite of that. So yes, it's work. It has to be.

But "struggle" gets romanticised by people who've confused suffering with seriousness. The best work I've made happened when the conditions were right and the tools were good and the people involved trusted each other. That's not painless. It's just not painful for the sake of it.

Rhythm, mostly. Whether this cut lands on the beat or just after it. Whether three seconds of silence will make the next shot hit harder. Whether the viewer needs a breath here or whether we should keep pushing.

It's closer to playing music than to writing. You feel when it's right. The intellectual justification comes afterwards, if anyone asks.

Fewer people doing more. Not because people become unnecessary, but because the tools let a small team match the output that used to need a department.

The difficult bit isn't the technology. It's the business model. A freelance editor with generative tools can produce work that would have required a studio. But the studio charged studio rates. The freelancer charges day rates. How does the freelancer capture the value of the output rather than billing for the hours? That's the question I spend most of my time on.

Because the job description doesn't exist yet and the serious-sounding alternatives are all lies.

"AI Creative Director" implies you're directing the AI. You're not. You're negotiating with it. "Generative Video Specialist" sounds like a job at a consultancy. "Prompt Engineer" is already a punchline.

Slopwrangler is honest. The raw output is slop. The job is wrangling it into something that serves the project. And it's memorable, which is what a job title is actually for.

Any musical instrument except a flute. Mainly guitar, piano, and bass, but the list also includes drums, mandolin, ukulele, harmonium, and a brief, ill-advised period on trumpet.

I played in several bands through my twenties and thirties. Still do, when schedules align. The musical training is more relevant to editing than people expect. Rhythm is rhythm.

I have two daughters, so the honest answer is: parenting. Beyond that, playing music, arguing about films, cooking badly, and reading about whatever I'm currently obsessed with, which at the moment is the history of printing and how every previous leap in reproduction technology followed the same pattern we're seeing now.

A state where my whole family is wealthy enough for me to spend all my time making things for fun and playing music with my friends.

There's no way to get there on a day rate. The gap between what I earn per day and what "wealthy" means is too large for linear accumulation. I need to find a seam where old skills and new tools combine to create disproportionate value.

Most of what I do now is oriented around finding that seam. The generative work, the lectures, the research, the tools. All of it is exploration with a commercial purpose.