Brian Cunningham on AI-Generated Documentary ‘Immortal Obsession: A History of Vampires’

Photo provided by Alex DiVincenzo

Brian Cunningham, creator of the completely AI-Generated Documentary, Immortal Obsession: A History of Vampires, spoke with Fandomize Media not only about creating this project, but also about his support of the WGA during the strike.

Innovative filmmaker Brian Cunningham utilized ChatGPT to write a compelling script, a video generator to create a lifelike host, an art generator to provide striking illustrations, and a music engine to compose a dynamic score. Cunningham’s human touch was used to refine the script and edit together all of the AI-produced assets.

Following the success of The UFO Chronicles: A History of Mysterious Sightings — the first documentary feature written, illustrated, and performed entirely by AI — ThoughyFly Films has harnessed the cutting-edge technology to create another film: Immortal Obsession: A History of Vampires. This documentary explores the classic monster’s captivating history, from their origins in ancient folklore to their enduring role in modern pop culture. 

Immortal Obsession is a thrilling example of how AI can take filmmaking to the next level,” explains Cunningham. “By using cutting-edge technology, we’ve created a film that’s both scary and sophisticated — just like vampires themselves.”

Check out our interview with Brian:

So, this is a completely AI-generated documentary. It’s kind of hard to wrap my head around that, so can you describe exactly what that means?

Brian: What it means is– so basically, here’s what the human did: I came up with kind of the concept of it, you know, I want to make a thing about vampires. And then fed that into ChatGPT with a really specific prompt, and then working with ChatGPT it generated every word of the script, if that makes sense. So we had it make an outline and then had it write each section and write transitions between the sections and then from there, all the images that are in it are generated images. The person that’s speaking is also generated just from the script. So we input the script, hit the go button, and that person comes out speaking. So, that was kind of the process. There’s still some human to it, you know, we’re not to the point yet where the AI can go and tell this program to do this and tell this program to do that. So basically, I’m kind of a slave to the AI, making sure all the prompts are working and making sure that everything kind of works together and comes together into one cohesive thing. Once it gets to the editing process, that is the tedious part where I have to basically do the little camera moves on all the images and the transitions and things like that.

So, just to make sure that I understand it, you kind of let the AI do its thing and then you kind of like, rein it in and organize it, right?

Brian: Yeah, it’s more like I’m kind of an editor, kind of a copy editor and then an editor when it comes to the images that it gives me as well, you know? So there’s still definitely a human touch in there. But it is written, illustrated and performed by AI. 

Very cool. The software, ChatGPT, what does that stand for?

Brian: You know what, I’m not even sure what GPT stands for. It’s a language learning model by OpenAI. It’s kind of exploded in the last couple of months. So with ChatGPT, you go in and you feed it a prompt, you say, write eight paragraphs about the Russian Revolution, and it will actually write those paragraphs. And if you say, write those paragraphs focusing on this historical figure, it’ll rewrite those paragraphs to function in that way. It’s kind of mind blowing how good it’s gotten. I’ve only been working with it for the last five months, I guess, and just the difference from when I started working with it till now is insane. Like it’s finally reaching the point where if you give it a decent prompt and tell it exactly what you want it to write, what it brings back is really hard to distinguish from what you would give a professional copywriter. It’s a little scary, but fun scary.

How long did it take you to put together this documentary?

Brian: This one took a couple of weeks. It takes about a day to write it and then it takes a couple of days to get all the images. That’s really the hard part because unfortunately, you still have to go into the image program and upscale the image, download it, import it into Premiere, all that stuff. All the boring stuff is what takes the time. And then from there, it’s just animating it. I also have an AI program that takes stock music and rewrites the stock music to fit the space. So like each section has its own piece of music that’s kind of custom written to that section. So I have to do that. But all told, as far as like, actual time sitting in front of a machine, it’s probably a full week of work to do one of the documentaries. Eventually I’m hoping that we can write some kind of algorithm that I can tell it, ‘hey, go prompt this image, download it, put it in the timeline and animate it.’ I don’t think we’re too far from that. So hopefully, the time for these kinds of things is going to go down significantly. 

Yeah, that’d be cool. For sure.

Brian: Yeah, and I don’t do it full time. It’s a week’s worth of work and I spread it out over two or three weeks. Usually I do 15-20 hours a week, just kind of whenever, you know, no hurry. There’s such a thing as doing too many and doing too much of this stuff and I want to make sure I’m kind of fresh to keep an eye on the AI and make sure it’s doing what it needs to do.

How did you first become interested in using AI? And then how did you come up with the idea to do an AI generated documentary? 

Brian: I mean, it was really kind of silly. I was on TikTok one night, bored and scrolling through Tik Tok. And there were just all these videos all of a sudden popping up about this thing called ChatGPT-3. Usually when you see stuff like that, on TikTok, you’re like, okay, well, this is some really dense programming language that you have to go in and install all these things and, you know, databases. But I wanted to look into it and when I did, ChatGPT-3 was just a web interface. You just typed in a prompt and then it would give you back text. And so from then I was like, well, that’s interesting. It would be really cool if I could get someone to read this. It would help me in my script writing. I’ve done traditional movies a lot before, I own a video production company, so we do a lot of script writing and things like that. And I was like, well, GPT can help with our script writing, maybe, you know, help us outline, help us proofread and things like that.

And then I was like, well, maybe we can get someone to read this script so we could use it for our commercial videos. And then the great TikTok algorithm fed me something about how you can input text and video back for speakers. I was like, there’s no way. That sounds too complicated, because we’ve all heard deep fakes, you know, like, hey, this is a deep fake of this person and I just kind of always assumed that these were like, really sophisticated programs running on server farms that people were doing. But then I found a subscription program where you just put in the text and I did a test where I just input the text for six or eight paragraphs and hit go. And what came back was shockingly realistic with the person standing on screen and talking to me. So that was kind of the point where I was like, holy crap, you could do an entire documentary about just about anything you wanted, because you don’t have to worry about who you’re going to get to speak or getting a camera crew to film it. You don’t have to worry about stock footage or licensing different types of footage because, you know, I’ve done some documentary work in the past and that’s the pain in the butt. You get all these still images and footage and you have to make sure it’s all licensed and correct. It’s like well, as long as we avoid any overt logos or anything like that, then we can probably prompt and get completely custom images. 

And then I was looking at it like this would be really inexpensive to do. So that’s when I was like, you know what, let’s just give it a shot and see if it turns out well, and that was our first documentary, which was The UFO Chronicles, which we put together really quickly trying to kind of learn what we could do with it. And I was still blown away with the quality of that. But the funny thing is this one was made a few months later and it’s leaps and bounds better than the first one. It’s kind of blowing my mind how fast these image generators and writing programs are progressing. So, that’s kind of how we ended up here.

How common would you say using AI in film is now?

Brian: It’s not terribly common to use it in the way that I’m using it. There’s AI, for instance,  there’s AI assisted programs like [Adobe] After Effects. So if you want to do an effects shot where you remove a picture or a painting from a wall or something, that will use a form of AI in order to look at it and figure out what’s the painting, what’s the wall. So AI is kind of working its way into individual tools, but I don’t think it’s yet made it kind of to the mainstream of filmmaking that I’ve seen. I do think that’s going to change in the next six months, though.

Just a quick clarification– what’s the difference between AI and something like CGI?

Brian: So, AI is a system where there is an intelligence to it. So, a system that would look at a situation and then run an algorithm or run its system in order to learn and to analyze, if that makes sense. At least that’s my understanding of it. So, you know, for instance, with the removal of the painting on the wall, you can just say, hey, here’s this painting, kind of roughly outline it, and then the AI will analyze each frame and say, well, that looks like a wall and that looks like a painting. It kind of has some level of intelligence to be able to figure that out. CG is a tool that’s run on– CG and AI can work together. Like an AI can create computer generated images, but traditional CG doesn’t, you know, anytime that there’s a need for that intelligence to kind of decipher or need this to look a certain way, it would need a programmer. It would need a person sitting in front of a machine to make those decisions. What we’re moving into is where you can give a really general prompt to the machine or the program and it can do it without needing your input or your feedback regarding each individual frame or, you know, how to do it. It kind of just takes the reins and runs with it.

So, with the WGA strike still going on, I have to ask– I mean, obviously, your documentary has proven that AI has merit in the film industry, and, obviously, so do the writers, so what should people remember about this discussion of AI versus a writer?

Brian: Well, I think that the situation I’m in is very unique, where I’m doing a documentary that’s extremely specific. It’s very low budget and the point of the documentary is to be very upfront with people about, ‘hey, we’re making this, yes, to give you something entertaining, but also to show you what this technology is capable of.’ I think it’s important for people to see how fast it’s moving in a real way so that we can have these conversations about, you know, ethics and how we can protect writers and how we can protect people’s creative property moving forward. I would not unleash this on a larger project, if that makes sense. If someone’s gonna say, ‘hey, we don’t need writers now because we have ChatGPT,’ I think that that’s insane, ridiculous and also offensive. 

What I’m doing is very much, it’s one of those things where this documentary wouldn’t exist if we couldn’t do it with AI because that was the whole point. But I think if you’re, you know, National Geographic making a documentary, and you’re using ChatGPT to write it, you’re really doing your audience and the writers that work for you a disservice, if that makes sense. Also, in the long term, this is moving so fast that it’s really hard to know where it’s going, but in the short term, ChatGPT is a tool. For what I’m doing, it’s a very specific task and it can do it great, but for what most writers are doing in the creative field, you still need a writer to be there to kind of guide it to come up with a story, something unique and original. Originality is still not its strong suit at all. I think what you’re going to see in the future is writers using it to help kind of kickstart their creative process, and to help fill in the gaps of the boring stuff, like helping them write pitches. I use it a lot for things like a pitch or a synopsis, you know, take this big piece of text and turn it into three sentences, you know, stuff like that. So, I think you’re gonna have writers that can use it as a tool similar to when Word Processing came out. It sped up the process so much and I think ChatGPT is going to be able to do something similar to that. But I do think there’s also a huge danger. There’s going to be a lot of corporate suits who think that they don’t need a writer and they can come in and write the next blockbuster movie. And I’m pretty sure it’s going to be terrible. At some point that’s going to happen, and it’ll probably be entertaining and all the wrong ways. 

I would agree.

Brian: I’m not saying that the technology is to the point where it can do everything. That’s kind of why I picked really specific documentaries, because it can give you a really good, interesting overview on a subject. So, for an hour and a half documentary about vampires, it’s great because it can lead you through that whole thing. For a three hour epic that’s trying to tell a really complicated story with character development, it’s not even close to being there yet.

That’s good to know. 

Brian: Yes, but I’m also kind of scared because it’s moving so quickly. Who knows where it’s going to be in a year or two? That’s the scary part and that’s why I think it’s important for people to stay up on it. We can’t kind of come up with the rules and legislation and, you know, the union agreements without knowing what it is and isn’t capable of and how quickly that’s going to change.

Right. So with all of that said, what’s up next for you?  

Brian: Yeah, so we’re starting to integrate some AI into kind of commercial work. For little web videos or promotional videos where they don’t have a $10,000 budget, this is great, you know, to bring in some of the AI to help do a little TikTok video or to help supplement things. So, we’re doing that on the commercial side. And on the creative side, I’m really enjoying doing these documentaries. It’s funny because I have an idea that’s like, I don’t know much about that, or I’m really passionate about that, and within a few days, I can get deep into it. It’s funny because I ask ChatGPT to write these scripts about things that I’m interested in and then it’s almost like I get to read a book about it while I’m doing it and learn more. There’s so many times where I’m like, oh, I didn’t know that about vampires. I’m really enjoying that. 

What I’m interested in is staying on top of what’s easily accessible to people without programming experience, if that makes sense. So, showing what these tools are capable of, you know, just for the average person on the street. I’m getting into some of the new text to video generators, which are fascinating. Everything I’ve done is text to stills and then we animate the stills, but text to video generators are going to be able to create stock footage from scratch. And then I want to see what ChatGPT gets good at as far as narrative. So my goal is just to keep making projects to showcase what AI is capable of so that everyone can really look at it and say, okay, ‘here’s where we are now, what does that mean for the industry and society as a whole?’ That’s why I’m really upfront with all these– I think the danger comes if someone’s making movies or creative projects with AI and not disclosing it. That’s where we get into this kind of danger zone of people trying to manipulate the audience. I’m always really upfront with like, ‘hey, this was made by AI half the reason this exists is so you can look at it and make your own judgments about the capability of the technology.’

Gotcha. For your next AI generated documentary, what topic would you like to tackle?

Brian: I actually just did a Buddhist documentary. I’m a Buddhist and I wanted to see how it would handle a totally different voice. I basically asked him to write a documentary in the voice of, you know, a mixture between like, a PBS documentary and a Buddhist text. I was blown away with how well it did that. That’s the next one that’s done and just waiting to come out. It’s really, really cool because I got to dig into really beautiful Buddhist imagery and play around with kind of the more fantastical elements of the art generation. And then after that, I think I’m gonna dig back into aliens a little bit, because there’s nothing better than a good AI-generated alien documentary at two in the morning.

If people want to stay up to date with your work, your documentaries and learn more about AI, where can we follow you?

Brian: Yeah, so I’m on Twitter @bjcunning. Also, you can look up ThoughtFly Films everywhere. And that handle is in just about everything. And then our YouTube page, where I put up a lot of the trailers and kind of updates is @thepostapocalypse. I’m a little spread out, I jumped into this without a huge eye toward branding, but that’s where we’re at.

Photo provided by Alex DiVincenzo
Sink your teeth into Immortal Obsession: A History of Vampires on Prime Video.
LEAVE A REPLY