15 years as a game dev. On this one, I didn't write a single line of code. Built it with @claudeai and a few other AI tools.
Clips from 2 finished chapters of my TPS game. First full mission drops Oct 9.
Be honest: would you have guessed this was built with AI? And knowing it was, would you still play it?
15 years in game dev taught me a few things worth passing on. Mostly AI here, but now and then I'll share general game dev tips. Follow if that's your thing 🎮 https://t.co/JuxkIlGWcL
Wow, this looks fantastic! I really admire you guys for making videos like this. I’m also fully integrating AI into my game development process, and I’m super happy with the results. As much as I’d love to share the details, I’m just too lazy to put together a video like this haha!
The games we've long wanted, which didn't exist because they'd cost several times more to make than they'd earn back, will soon become possible. As will the games developers wanted to build, but couldn't with the resources available. It'll be a new golden age of gaming.
@jurlycat Sorry, is this yours or another dev's? Love the bond mechanic. Been wanting to build a 3D survivors-like for a while, would love to follow along.
I've always loved the conversations that happen while you're just walking around in a game. Not cutscenes, just characters talking, reacting, being themselves. It makes the world feel alive.
So I wanted that in my game too, on an indie budget. @ElevenLabs made it possible.
Here's how it works on my side:
Every character has a bible: who they are, how they look, how they talk. Every voiced line in the script comes with a bit of context about what's happening in that moment.
Claude reads the situation and the line, figures out the tone and delivery, then calls the ElevenLabs API.
If a character hasn't been cast yet, it runs a little casting round first. I get a few voice candidates, listen, and pick the one that feels right. For characters already cast, it gives me 2 or more takes per line. Sometimes I mix takes line by line until it sounds right.
After the first time, I turned the whole thing into a reusable workflow. Now I just say "do the voice-over for mission X" and Claude handles the rest and sends me the takes to review.
Here are a few of the lines from the game 🎧
@Shawn0578226690@Belbizaz@threejs And yes, console have browser then how about if that browser not allow to run js? And the control and etc, how to solve it?
@Shawn0578226690@Belbizaz@threejs It’s not as simple as you think. Game development isn’t just about getting something to run somewhere. There’s optimization, security, and a whole lot more. My 15 years of experience in game development would count for nothing if it were all that simple.
@icreatelife Hello Kris, this is one of several projects I’m working on (you can check out more on my page). Firefly would be a fantastic addition to the project’s tech stack, should the opportunity arise!
https://t.co/NpVuRUrP97
15 years as a game dev. On this one, I didn't write a single line of code. Built it with @claudeai and a few other AI tools.
Clips from 2 finished chapters of my TPS game. First full mission drops Oct 9.
Be honest: would you have guessed this was built with AI? And knowing it was, would you still play it?
@migueldelaluz1@null_assets That is possible as long as you want it. This is a small experiment of mine, specifically referencing the style of Chaos Zero Nightmare: https://t.co/vbbrYNa4Ux
Vibe coders are building every kind of game right now.
But one genre I almost never see: the anime gacha card battler. The genre with flashy skill and ultimates, cinematic cut-ins, and VFX on every card.
So while my main AI game keeps moving, I spun up a side project to test exactly that. Meet "STARFALL demo" 👇
Everything you see is AI-made. I didn't write code or touch the image, animations by hand:
• @ChatGPT Codex (Astra + Sol 6.1) wrote the game
• Godot engine
• @Grok Imagine (free tier) for character images, animation, ultimate and effects
The numbers:
• 1 prompt + 1 reference combat clip → 6h of agent work
• ~4 rounds of feedback → 4h more
• ~10h total → a full playable combat encounter: 3 heroes, with skills and cinematic ultimates, cards, boss turns
This was a quick test bed for anime card battle animation tricks, built on a budget stack with speed over polish. My production stack goes much further: stronger models, higher-res assets, and a proper QA and polish phase.
Gacha anime games are notoriously expensive to make. What happens when one dev with AI can prototype one in a day?
If I turned this into a full game, would you play it? 👀
It depends on many factors. These are just initial, small-scale figures. Rates often appear high with a small sample size, but they will gradually adjust to reflect the true reality over a sufficiently long period and with a large enough volume of traffic. You will never see such high metrics once your traffic reaches a truly significant scale. A high click-through rate (CTR) combined with low traffic doesn't really indicate effectiveness; you need to look at other metrics such as wishlist additions to gauge the actual conversion rate.
If you’ve had a good start, keep building on that momentum. These figures come from three of my own games, the last image is from a "coming soon" game that recently launched its store page and boasts excellent performance. Just to show you that I’m not speaking without evidence.
Honestly better than I expected for free tier! From what I saw, Codex generated a still portrait of each character first, then used those as the reference/start frame for every video instead of pure text prompts. That kept consistency mostly under control. A bit of drift on small details, but a quick re-gen fixed it 😄
Vibe coders are building every kind of game right now.
But one genre I almost never see: the anime gacha card battler. The genre with flashy skill and ultimates, cinematic cut-ins, and VFX on every card.
So while my main AI game keeps moving, I spun up a side project to test exactly that. Meet "STARFALL demo" 👇
Everything you see is AI-made. I didn't write code or touch the image, animations by hand:
• @ChatGPT Codex (Astra + Sol 6.1) wrote the game
• Godot engine
• @Grok Imagine (free tier) for character images, animation, ultimate and effects
The numbers:
• 1 prompt + 1 reference combat clip → 6h of agent work
• ~4 rounds of feedback → 4h more
• ~10h total → a full playable combat encounter: 3 heroes, with skills and cinematic ultimates, cards, boss turns
This was a quick test bed for anime card battle animation tricks, built on a budget stack with speed over polish. My production stack goes much further: stronger models, higher-res assets, and a proper QA and polish phase.
Gacha anime games are notoriously expensive to make. What happens when one dev with AI can prototype one in a day?
If I turned this into a full game, would you play it? 👀
@Adolphite Agreed! Since this is just a test version, I’ve only set it up to include the essential elements for a single, complete battle. For the final product, much more detailed refinement will certainly be required.
Yes, I build games. But I also make things just because I love them, and short AI video clips are one of those.
This one is a little Y2K bedroom moment: a girl, an old digital camera, a mirror selfie, and a printed photo at night. The kind of quiet nostalgia I grew up with.
How I made it:
• 4 keyframes generated with AI, same character in every shot
• 3 video segments in Seedance using first + last frame, so each shot flows into the next
• Stitched in CapCut, trimming the duplicate frames at every join so there's no stutter
15 years of game dev taught me one thing: tools change, but storytelling stays the same. AI just lets one person do what used to take a team.
What era should I bring back next? 👇
hood warfare 2 uses epic's asset store for its models and they honestly may have even used ai in the coding.
currently with 10k+ players, i wonder where the people who say "using asset store items is slop" and "ai code is slop" are at?
The game is $5, and 10k players means they made already 50k.