So, I have a lot of thoughts about how AI (especially image and video generation) can be used intelligently and creatively and how its potential isn’t really being tapped because the ‘pro’ camp seem to think it can and should replace meaningful creative endeavour while the ‘anti’ camp think it can’t and shouldn’t and therefore nobody should even touch it ever. I wanted to look at what could be accomplished when AI tools were combined with clear, human-developed visions and plans. I eventually want to use this concept to create a full, original short film, but first I needed Proof of Concept and a chance to familiarise myself with both the technology and with basic video and cinematic storytelling techniques. I’m normally a prose writer, not a film-maker (as I’m sure you can tell by the above rough-and-ready nonsense), so I couldn’t take either my familiarity or competence as a given. Luckily, I have two recurring characters in my narrative roster who most of my regular readers will be familiar with: the functionally-immortal superbeing Crescendo and the Last of the Wizard Kings, Lacurio. Contriving a reason for them to beat seven shades of shit out of each other across dimenions (to pounding punk rock music, no less) gave me a chance to acclimate myself to the medium without having to worry about conveying a complex, multiphasal story. Plus, I got to do a little Ketarch cameo at the end for those of you who read my novel Agent of Nowhere. So that was fun.
Anyway, let’s break down what’s going on here a little. Obviously, the AI is doing a lot of the heavy lifting on the technical end. The references images that keep the characters (relatively) consistent across scenes, the backgrounds into which their integrated and the footage itself is all AI-generated. On the surface, it might seem like my authorship is questionable, as so much of the basic meat-and-potatoes filmmaking was automated. However, hopefully, you can see that there is a lot of human direction to the above video — and not just humanity, but my personality specifically. It’s in the way our characters dress; how they move; how they fight; the rich, hypermaximalist aesthetic of the worlds through which they battle; the manner in which their personalities (of which the AI has no concept) shine through despite the absence of dialogue. That’s because a lot of thought was put into all of the above long before I started making the video. Indeed, eagle-eyed viewers will notice that many of the locales appeared in previous Crescendo or Lacurio stories, either just in the text or as accompanying images. However, there’s one scene that I think perfectly encampsulates the way this just wouldn’t have been possible without a real human intelligence directing proceedings (spoilers ahead if you’re reading this before watching the video).
Okay, so Lacurio has just unleashed a tremendous burst of fire magic, blowing up the gothic temple in which he and Crescendo were duking it out and they’ve been thrown into the street outside by the force of the blast. The moon is bleeding, but that’s an incidental detail and not relevant to our ‘heroes’, who have both seen much weirder things. They are, however, somewhat singed and in need of respite. They limp to the nearest pub, pause to sit and drink a glass of wine… and then get right back to fighting. It’s a very deliberate moment of humour, yes, but it also shows off their personalities more clearly than anything else in the long, over-elaborate fight-scene. When the time comes to rejoin battle, Lacurio stands up purposefully, throws his glass of wine aside to shatter, unheeded on the floor, and prepares to resume noble pugilism. Crescendo, still seated, draws a gun and shoots him. On the one hand, the flamboyant egotist, passionate but posing in the middle of a brutal, life-and-death struggle. On the other, the cunningly-disguised noir anti-hero, acting pragmatically and unromantically to get the job done because he’d rather be literally anywhere else right now. Both, however, are ultimately gentlemen, and willing to put aside their differences for a moment’s truce. They don’t even need more than a single significant look to establish this. AI couldn’t do that on its own. Only someone who knows these characters could. I might not be hand-animating each scene — sadly, that’s just not a skill I have — but here, I’m serving a similar role to a film director. I’m not doing the acting, but I’m managing and guiding the actors.
You see similar expressions of the protagonists’ personalities all throughout the piece. They react to the letters that trigger the fight differently, for example. Crescendo takes his letter from Enigma (who has a brief appearance at the start of the video), reads it calmly and doesn’t seem to react, internalising his feelings. Lacurio is visibly moved to alarm, crunches his into a wad of paper and immediately opens a portal between realities with a needlessly exaggerated gesture. When they meet, Lacurio is angry, acusing. Crescendo, who is hundreds of years old and just doesn’t want the fight happening in his living room, bows, lowering the temperature of the interaction enough to move the battle outside the house and onto a formal footing. Then, throughout the fight, Crescendo is trying to have a normal dust-up with fists and weapons. He’s a superb combatant, and that’s what he understands. It’s Lacurio who keeps deliberately changing the venue and nature of the match: opening doors between worlds; summoning tidal waves; creating pocket-dimensions from playing cards; doing everything he can to magically unsettle and wrongfoot his opponent. Magic, after all, is his metier, not guns and martial arts moves, and this shows throughout.
Even Ketarch’s cameo at the end is specifically calculated — right down the fourth wall break — to show that this is man in complete control; affable yet sinister; one step ahead of everyone else; both reassuringly calm and chillingly unmoved by the violence he caused. It’s not something an AI is doing without very specific, careful, planned human management.
Then there’s the scene transitions. The fight had to flow smoothly from one setting to the next without the transition feeling forced or nonsensical. If an alternate universe shatters, our heroes emerge from a shattering mirror. If a building interior explodes, we cut to the exterior to show Crescendo and Lacurio being thrown out of the blast. If a character is drowning in magma, the next scene is the volcano erupting and sending them flying. All that sounds simple, but in practice, it takes a surprising amount of planning and effort, not to mention non-AI, old-fashioned video editing. It’s that human element again, keeping the project from feeling dead inside.
I guess what I’m driving at here is that AI isn’t inherently a bad creative tool, but too many people abuse it and expect it to do the work of creative labour for them. The reality is that, like any other system, it has uses, it has limitations, and it ultimately requires a passion for invention and storytelling to be wielded effectively. Hopefully, what I’ve shown you is that this still nascent technology isn’t a curse or a blight in itself. It’s been misused and treated as a crutch, yes, but in the right hands, it can help creators realise some pretty cool projects that would otherwise be out of reach without the money to hire actors, build sets and rent green-screens. This one was a fun one to muck about with, but it was only a first effort. Expect the next one to have sound effects, more varied action and a plot more complicated than “two very well dressed men punch each other a lot”.
PS. The song I used was Bombshell Blackout by Tragic Eight Ball… who are just me in another AI flavoured hat. I wrote the lyrics, coded the rhythm into the line-breaks and punctuation, decided genre, tempo and singing style, then got software to perform it and fill in the instruments and melody. I think it fucking rocks, but mileage may vary.
