Google’s new video generator is out in full force, and I don’t know about you, but my feeds are full of examples of what it can (and sometimes can’t) do. I’ve already covered how it’s deepfaking smooth-brain YouTube content, half-baked Michael Bay bulls**t, and absurd pharma ads, but there’s another category I haven’t talked about yet, and it’s potentially a lot more impactful than the former uses. Sorry, gamers, but Veo 3 (and likely other video generators) might be on a collision course with your AAA games next.
Veo 3 does video games really well. It’s like you can explore new worlds already.
Prompts are all variations of:
> a third-person open world video game walking around…
> an fps video game in/on a… pic.twitter.com/bpDGaKNU55— fofr (@fofrAI) May 22, 2025
One area that Veo 3 is surprisingly functional at is generating fake gameplay footage of nonexistent AAA games, which—if you’ll allow me to turn my brain off for a moment—is pretty entertaining. Above are the results of a prompt that asks Veo 3 to generate a third-person open-world game in different settings, including some realistic Grand Theft Auto-style generations, some sci-fi, some fantasy, and one that looks an awful lot like The Last of Us. Copyright concerns aside, it’s pretty neat!
It makes sense, given the never-ending stream of gaming content that Google might have at its disposal through YouTube, that Veo 3 would be able to spit out some kind of facsimile of a AAA game. That’s not the part that really intrigues me; it’s that people are already taking those generations and running with them. Below is an example from a 3D designer who goes by Lovis Odin on X demonstrating how Veo 3’s outputs can be integrated into a workflow that takes the AI video generator’s text-to-video powers and makes them even more granular.
Google Veo3 creates beautiful base videos, but what if that’s not enough?
I built a @ComfyUI workflow that takes it further:New structure with Flux (LoRA arch)
Turned into 3D with Hunyuan3d 2
Integrated + relight via Flux, Controlnet, Denoise and Redux
Finalized the… pic.twitter.com/pAfu2qVU34
— Lovis Odin (@OdinLovis) May 27, 2025
Using a few other tools in tandem with Veo 3, Odin takes an initial video prompt and then adds a structure that is rendered into 3D and is then fully customizable. Sure, Veo 3 can add objects on its own, but as Odin notes, what separates this workflow from your typical text-to-video prompt is the granularity. As good as text-based prompts can be, they’re also kind of a huge pain in the ass if you want something specific, which makes them less ideal for real, professional workflows.
Odin’s workflow isn’t solely focused on games, to be clear—this is the type of thing that could be applied to really any generative video—but it feels especially practical for game development. Think about it: you have an idea, you generate the look with Veo 3, customize it with other 3D tools, and then (though this part isn’t included in Odin’s workflow) you bring it to life with AI-generated code. Obviously, this last part is the biggest piece of the puzzle, and generative AI isn’t quite at the level of being able to properly code AAA games yet, but that doesn’t mean we aren’t trending in that direction.
People are already well along on their journey to “vibe coding” simple games, but something tells me that’s just the beginning. Google itself already broached the idea of generatively coded games back in 2023, stating in a blog post on generative AI and gaming that “creating content is one of—if not the largest—expenses that games can incur.” The blog also said, “According to the UK’s CMA, blockbuster games can have development budgets well over $100 million… AI allows developers to leverage this new technology in a way which is respectful of intellectual property while protecting one’s own proprietary data.”
To summarize Google, if I may, game development is expensive and time-consuming, and they see big potential in reducing the labor costs of games, copyright be damned. Is that surprising? No, not at all. AI, if we’re to believe the hype, appears to be a labor-reducing force across the board in lots of industries, games included. What is surprising, however, is that the progress in that direction might be coming faster than we think, and if it’s really here, that could be bad news for the real, extremely overworked people of the games industry and potentially bad news for games themselves.
As vast as the potential for augmentative AI may be, the potential for AI slop is just as high, and unfortunately, there’s nothing special about the games industry that would prevent that same slop-like ripple from taking effect. I’ve learned not to sound the alarms too early—things happen in the tech world, and they happen fast. Maybe in the not-so-distant future, we’ll wake up and realize this whole AI thing was just a passing fad, but if there’s one thing that’s for sure, it’s that we’ll likely find that out the hard way through a lot of Fortnite plagiarism.
