🔦 Creator Spotlight: Kavan The Kid
Directing with Machines: AI, Fan Films, and the Studio of the Future
This story features highlights from my conversation with LA-based filmmaker and Phantom X co-founder Kavan Cardoza, aka Kavan the Kid. I saw Kavan present at AI on the Lot in LA this past May in a forum spotlighting emerging AI studios. I left his talk feeling even more wonder about the future of creativity and technology and all of our places in it. When I started this project, I knew I wanted to talk more with Kavan about his journey.
From viral fan films to his landmark short Echo Hunter, a hybrid film that combines AI-generated visuals with a fully SAG-AFTRA cast, Kavan is pushing the boundaries of how we make movies without losing sight of why we make them. His work is defining what a tech-forward studio of the future looks like, blending human imagination, real actor performances, and a strong ethical compass in a machine-driven pipeline. I loved chatting with Kavan, and hope you enjoy the conversation.
This article presents excerpts from our longer conversation. Some responses have been lightly edited for clarity and brevity.
📺 Watch the full conversation on YouTube or 🎧 Listen on Spotify
Kavan the Kid: The reel before the feature
🎬 The Fan Films to Phantom X: Kavan’s path runs through rap music videos, viral AI fan film experiments (Power Rangers, Star Wars, and GI Joe to name a few), and now Phantom X, which bridges Hollywood systems and next-gen tools.
🎥 Setting the SAG Standard: With Echo Hunter, he directed one of the first AI-driven shorts to feature a fully SAG-AFTRA cast (including Gedeon Berkhard of Inglorious Basterds).
🛠️ Hybrid Workflow Architect: Masters a custom stack of generative tools while always centering human performance and vision. Kavan treats AI like a camera more than a creator.
📈 Setting Ethical Standards: Pioneers artist-first principles including 0% IP ownership over what his creative collaborators produce.
"Somewhere out there in the world is the next Tarantino… if they have a solid story and a drive to tell it, I think we're just going to see an entirely new wave of filmmakers emerge."
Interview Excerpts:
Siddhi: What was your moment where you realized AI could change your filmmaking practice?
Kavan: Literally the first time I use Midjourney. Even though it was just a blob, I generated Deadpool and the concept of him was there, like the shape roughly of his head and you know, the black parts right here and the little white slits for the eyes and then the red…nothing of what he actually resembled, but it's there in the mess… I've always said with tech it's not a matter of if, it's a matter of when, and so I knew at that moment this is going to be really crazy... It is definitely the future. There's no denying that... it's only going to get better. And at the pace that it's moving, it's going to be here in a mainstream level quicker than I think most people are expecting.
Siddhi: What was the leap from those early experiments to making something like Echo Hunter?
Kavan: With every long form — and by that I mean something that's longer than five minutes — project we've done, we've tried to just internally for ourselves raise the bar of what we can do and what we can achieve ... I also didn't want to start expressing my own ideas until I felt the tools were at a point that I could make something like an Echo Hunter... when you were doing original ideas in the earlier days of AI, it was very hard to get that consistency. Whereas if you're doing existing characters the machine's clearly been trained on, it was a lot easier to get the consistency.
And I had this Echo Hunter idea for a long time and I felt like it was going to be good for AI because the story revolved around the characters having masks, which again, then hides some of the uncanny valley stuff of characters talking. But the masks are actually like an actual element to the story because the characters can't see their faces for the sake of their memories coming back.
Siddhi: What did it take to get SAG approval?
Kavan: There are several elements that we went back and forth on, we submitted it as being classified as a film. And they came back and had some more questions and then it was getting switched over to animation. And we were like it's not technically an animation because we are going to be shooting actors that aren't wearing masks, we are going to be shooting their faces and taking a live performance capture and implementing that using Runway's Act 1. So those facial performances are their facial performances. It's not on the same quality level or financial level, but how is that any different than what Avatar does? They do the exact same thing. It's a facial performance capture. And so then we ended up coming to a new thing called a new media deal. And in the structure of the deal were the rates, thankfully the entire cast worked with us within our budget constraints. But we had a higher SAG minimum than what would be in a traditional film. And then we had to agree that once the film was complete, all models that we had trained on the actors would be deleted. As well as outside of what you see in those 22 minutes. We can't create anything else with the talent.
Siddhi: Walk me through the production pipeline. What did that hybrid workflow actually look like?
Kavan: The actual production pipeline was scripting, casting, SAG approval, and shooting once we had SAG approval. And funny thing is prior to the SAG approval, I'm able to still work on the film, because it's AI. It’s how I even got the actors interested, I had the first minute and a half of the film done and I was sending it to them. Like, this is what it looks like. And they were like, whoa, this is awesome. I had a lot of like placeholders of my terrible acting skills in there, just so I could be editing and moving along. And then once we had SAG approval, that meant we had the green light to actually now go and shoot our talent. For Gedeon, he was based in Germany. We had him set up his iPhone ... and then on his laptop, we were zooming with him to direct him.
With Arcana, what I love about their model particularly is ... you just need anywhere like on the low end, 10 photos. I think we were doing about 35 photos of the talent, a mixture of photos that we shot on set when we recorded them and like their headshots ... You train the model and then you make some slight adjustments and then essentially what is really cool is you can either use Arcana's image generator or you can use Midjourney or Imagen or Whisk or Flux or you know any of these tools and then you can take the image over to Arcana and it's like on a paintbrush you just paint over the face, click on the model, set your percentage... then boom it's the person ... we were taking a live performance capture and implementing that using Runway's Act 1. So those facial performances are their facial performances... I did all the sound design myself. ... There weren’t AI sounds."
Siddhi: You opted for real actor voices and a traditional composer. Why?
Kavan: I wanted a very specific sound when it came to the music and as funny as it is, I was always in the music industry…but I didn't make music, I was on the visual side of it. So I'm not the best at the AI music tools. But with this, I really wanted it to be fine tuned. And I wanted it to actually have the rising and falling actions of traditional filmmaking to where when the intense moment's happening, it's not like I'm trying to cut a music video to that part of the music. It's like, no, if I want it there, it's gonna be there. But we did it in a very not traditional way. He (composer) would essentially send me a very stripped down track. I would be editing so we could move at a quicker pace. And so then I'd hit him back and be like, “hey man, all right on, track three at 36 seconds in, I need it to rise and I need it to be that way until like 42 seconds into the track. Then it needs to resolve itself.” And we would just would go back and forth like that for hours.
Siddhi: When you talk to a lot of studio folks, the thing that is the most scary right now to a lot of people is what the teenager in their bedroom capable of making with access to this technology. That democratization might be the single biggest pivot in creative history towards lowering the barrier to entry for telling great stories that I can remember.
Kavan: It's not about whether you make things faster or cheaper. To me, somewhere out there in the world is the next Tarantino… all the legendary directors and writers that we've come to know through the current hundred years of filmmaking are out there, but might not have the means or the connections or capabilities. And this could be anywhere in the world. And now they're going to have these things that if they have a solid story and a drive to tell it, I think we're just going to see an entirely new wave of filmmakers emerge… the studios can start noticing like, wow, this person's really talented. And then giving them a deal to then set them up, giving them the means that they might not otherwise have had because they never would have been noticed in the first place.
Siddhi: I talk a lot about the Creative Firewall™, the line between what the machine can generate and what needs to stay human. Where is that line for you?
Kavan: The storytelling comes from you… the idea is human made. Once that part is removed, I don't really know how much human-ness there is... when I look at AI, to me, it's kind of no different than the camera. It's just a different camera that I'm using. But someone might make the argument, well, what about set designer and costume designer and things like that? Well, I technically could have those people sitting here next to me and prompting, but the thing is, I'm coming up with those things. In Echo Hunter, Garak ends up with a black helmet and a long brown trench coat and the tan turtleneck underneath. That was my idea. I told it: this is how I want him to look down to the detail. And when you're describing the world and everything, you're telling it in detail… so in a sense, well… I was the set designer. I was the costume designer because I'm the one feeding all the information. It's just the machine is actualizing it.
Siddhi: What do you say to people who believe AI can never create emotion? Or the people who think that machines can do it better?
Kavan: Someone at one of the AI companies told me that they they’ve gotten better prompts from the machines than they have from real actors. And I was like, who are these real actors you're using? Because if you're using just anyone, there could be good filmmakers and bad filmmakers. There could be good actors and bad actors. But if the actor is talented and is good, these are professionals at their craft and have done this their whole lives and have worked to master their craft, they're going to give you phenomenal performances.
Siddhi: What’s your dream tool request for the companies building these models?
Kavan: Let me say two… first one is Runway, please update Act One! It's been like seven or eight months without an update. I would love to see an improvement there for those facial performance captures…but if I could have a dream tool, it would be a full body mocap of an actor. I want the whole body and then be able to generate an image in like Midjourney or Flux. And take that image and essentially skin the character with what's in the image and then have them do whatever the actor does…there are so many nuances that actors can do, very little subtle things that can bring you in. And you hear all the stories of all the movies where the actor improvises. And then something happens that was never expected. Well AI is not going to do that because it's going to do what's based on the prompt, period.
Siddhi: For creators who are curious but nervous about the generative era we’re in, what’s your advice on where to start?
Kavan: If you're just getting started and maybe you don't know if you want to go all in on this, because it can get expensive really quick, especially if you're not part of the Creative Partner programs, I always have the same two tools and I've said it from the beginning. I would use Midjourney for your images to start and for video I would use MiniMax. I think MiniMax is the most cost effective but yet delivers the best results, especially with their newest model. It's on par with Veo. But it's so much cheaper. So I think it's a great entry point for someone to not break their bank. You can essentially spend roughly about a hundred dollars a month and start creating and messing around telling stories and then if you want to go a little bit further, you know with wanting the characters to talk, I would stick with Runway because I like Act One. But if you don't want to use actors, both Hedra and HeyGen are awesome. And then ElevenLabs for character voices, the new text models are actually really good where you can capture the inflections.
Kavan’s vision is both practical and visionary. It’s an incredible example of what happens when the director’s eye stays sharp, even as the tools evolve. His work reminds us that machines don’t make meaning, humans do.
✉️ If this reminded you that directing still shapes the soul of a story, even in an AI-powered world, hit subscribe and send this to someone daring to dream up their next film or creative project.