Givros
@givros • 3,287 subscribers
Build your next game from one prompt ⚡ Lifetime access to prompt 👇 https://t.co/9ZJpW3UDFD Open to Collabs 📩
Videos
GPT-5.6 Pro is insane. It didn’t just beat Claude 5, Fable and Mythos. It crushed them. One prompt, just 5 minutes. And voila: A full low-poly Three.js island game with water, beach bar, boat, shark, fish, whales, sports area, NPCs and Ibiza vibes This is getting ridiculous. That is what I would write if I was not honest with you. That is what I would post if I only wanted a viral clickbait tweet. And this is exactly why you should always be careful with what you read on social media, especially on X. There is no GPT-5.6 yet. This prototype was made with GPT-5.5, in 6 iterations, after more than 40 minutes of back-and-forth with Codex. Yes, it started with one prompt, and I will share that prompt below. But the final result came from many corrections, adjustments, fixes, visual tweaks, Three.js issues, camera changes, object placement improvements, and a lot of manual direction. That is the real workflow. I share build in public content here to show what you can actually do with AI, but also to show that there is no magic dust. AI is powerful. Very powerful. But the difference between a cool demo and a real usable prototype is still iteration, feedback, taste, debugging, and direction. Yes, I am also waiting for GPT-5.6. It will probably arrive next week, and when it does, we will test it here properly, in public, with real prompts, real iterations, and real results.
Givros62,031 views • 2 months ago
Codex is unusable. I generated one image with GPT Image 2.0, gave it to Codex, and used one simple prompt: “Create me a web game from this image.” You give it a simple image like this, and suddenly you realize you can build: • a voxel map • an editor • tile animations • a playable browser prototype This is getting ridiculous.
Givros83,376 views • 4 months ago
After testing img2threejs by Nick, I used the exact same Canyon gravel bike image with Blender MCP. The difference is pretty clear: Blender produced a much more detailed and faithful reconstruction, especially around the frame geometry, cockpit, drivetrain and braking system. img2threejs remains incredibly fast for generating a procedural Three.js result directly from an image, while Blender MCP offers much more control and precision when you want to push the reconstruction further. Same reference image, two completely different workflows. Which result do you prefer?
Givros31,979 views • 1 month ago
Codex isn’t just a coding assistant anymore. It’s becoming a full AI game-dev workspace. 🦫🎮 For this project, Codex handled the workflow by using: 🖼️ GPT Image 2.0 for the visual direction, cards, enemies, and UI assets 🧠 GPT-5.5 for gameplay structure, systems, and iteration 💻 Codex to turn all of that into a real project From cozy capybara shrine images to a real turn-based card battle game. The 2026 AI game dev workflow is getting ridiculous. 🚀 Follow if you’re building games with AI. Want the full prompt I used for this? Comment “CAPYBARA” and I’ll share it.
Givros71,070 views • 4 months ago
Codex is a monster with 2D games. But what is it worth with 3D and Three.js? 👀 Codex + GPT Image 2.0 + a crazy prompt = • playable 3D forest scene • humanoid character fully made of leaves 🍃 • dense foliage covering the whole body • third-person movement • wind animation on every leaf • SPACE transformation into a flying leaf swarm • leaf trail while moving • runs directly in the browser AI doesn’t just help you code faster. It lets you test ideas that would normally feel too complex to even start. Comment “leaf” and I’ll share the prompt. #ThreeJS #WebGL #Codex #GPT5 #AIAssistedDev
Givros60,844 views • 4 months ago
How do you make a vibe-coded world feel truly alive instead of looking like a few assets randomly scattered across an empty map? I keep running into this problem. The AI can generate the terrain and each asset individually, but it struggles with placement, density and composition. A forest becomes ten isolated trees. A town becomes three buildings beside a road. The assets exist, but the environment never feels naturally constructed. It made me think about last year, when I opened X and discovered the insane project @levelsio was building: a multiplayer flight game running entirely in the browser, with a sponsorship system where brands could buy their own advertising blimps. The entire project was vibe-coded. Watching such an ambitious idea become a real multiplayer game was incredible, and it was exactly the kind of project that pushed me to go 100% into vibe coding. Now Astra is emerging as OpenAI’s next major model family (whether it eventually ships as GPT‑5.7, GPT‑6 or something else) and I’m ridiculously excited to discover what this new generation will let us build. So here’s a little nod to that inspiration with my own project where you can pilot a plane. And yes, I absolutely love Disney’s Planes 😂 How do you handle environmental placement in your projects? Better prompts, procedural generation, specialized tools, or manual level design afterward?
Givros17,204 views • 29 days ago
I gave GPT-5.6 Sol Ultra a full set of character references: front T-pose, four-side turnaround, idle, walk and attack poses. The goal was not just to generate a nice mage image. It had to rebuild the character in Blender, rig it, create 3 animations, export a real GLB and then build a complete Three.js showcase around it. The full goal was completed in 2h11. The pipeline works: the model loads in the browser, the camera can orbit around it, Idle / Walk / Attack are playable, and the page displays the real mesh, material and animation data. But the before/after also shows exactly where image-to-3D still struggles. The overall identity is there, but the proportions became much more chibi and blocky than the original reference. The hat lost its clean folded silhouette and turned into a striped curved tube. The face and eyes are too large and too bright, while the original had a darker hood with much smaller cyan eyes. The robe layers, shoulder shapes and front ornaments were also simplified too aggressively. The back is clearly the weakest part: the cape became a large smooth cone, most of the costume structure disappeared, and the head turns into a glowing purple sphere instead of preserving the dark hooded silhouette. Some gold trims are uneven, the hands are still rough, several parts intersect, and the materials look much more saturated and plastic than the reference. So no, this is not a perfect “one prompt 3D character” result. But going from a few generated references to a rigged, animated GLB running inside a custom web showcase in 2h11 is already a pretty interesting baseline. Now the real work starts: fixing the silhouette, rebuilding the back, cleaning the materials and improving the animations.
Givros24,498 views • 1 month ago
This is the part people don’t show enough about AI game dev on X. Behind every “Fable 5 made me a full game in one prompt” post, there’s usually a lot of bullshit. I gave AI a strong /goal: optimized Three.js grass, chunks, LOD, shader wind. It produced something. It understood part of the direction. But it still missed. That’s the real workflow. A prompt can give you a starting point. It doesn’t magically give you taste, debugging, optimization, and iteration. You can test the /goal below.
Givros26,603 views • 2 months ago
This is the best 2026 AI workflow for turning game ideas into playable prototypes: GPT Image 2.0 → reference image + tileset Codex + GPT-5.5 → my custom prompt Result → playable browser prototype I generated the Cozy Island Farm concept, turned the same visual direction into clean technical assets, then used one complete prompt to make it playable. Comment “Cozy Island Farm” and I’ll send you my custom prompt.
Givros32,683 views • 3 months ago
Claude Opus 5 just one-shotted this FPS. Less than 1 hour from prompt to playable demo. Every weapon, enemy, mechanic and environment was generated from scratch. Prompt below 👇 There is no prompt below, because the first section is exactly the kind of post I’m tired of seeing every time a new model drops. A flashy demo, “one-shot” in the caption, “built in under an hour”… anything to farm impressions while conveniently hiding the actual process. If it was really made from a single prompt, show the prompt. That’s what I do on this account. When I claim something is a one-prompt prototype built with GPT-5.6 Sol Ultra and Codex, I share the prompt publicly so anyone can verify it. AI is already making game prototyping incredible. We don’t need fake “one-shot” stories to prove it. Original footage:
Givros14,794 views • 1 month ago
The character pipeline is moving forward: animations are now working. I lost some time because Codex occasionally switched GPT-5.6 Sol back to Light/Medium by default, and the difference in task execution was immediately noticeable. After switching back to GPT-5.6 Sol Ultra, progress was much smoother. Using animations from Mixamo, the character can now idle, walk, run and jump. Next: shoes, skin variations and different body types. Then I need to decide what this project should become: • A character generator website? • A reusable Codex skill? • An open-source GitHub repo? What would be the most useful?
Givros16,804 views • 1 month ago
I want this exact kind of game. Simple controls, satisfying combat, constant progression and tons of upgrades. But without the gacha systems, the 15 buttons begging to be clicked, the endless pop-ups, the fake rewards and an ad every 30 seconds. Just let me play the game😭
Givros12,299 views • 1 month ago
I tested my Blender character creation skill with Codex + GPT-5.6 Sol Ultra. The full prompt: “Create a Three.js game where the player explores different biomes. Create the character and animations with skill-blender-character.” Codex built the world, created the low-poly character in Blender, generated the outfit and integrated the animations into the game. One prompt, one custom skill, one playable prototype.
Givros10,599 views • 1 month ago
“Use the reference image named farm_voxel.png as the main visual guide. Create a playable-looking low-poly voxel agricultural farm diorama" This is the workflow I keep pushing: 1) GPT Image 2.0 for the visual base. 2) Codex with imagegen + GPT-5.5 for the build. 3) Three.js for the prototype. result: 🎬 Prompt below 👇
Givros17,281 views • 3 months ago
I rebuilt the same 2D survival base defense prototype. But this time, I used Agent Sprite Forge of 0xFunky with Codex. Instead of relying on rough placeholder assets, the workflow can generate 2D sprite sheets, maps, transparent PNG frames, props, GIFs and metadata directly from prompts. And honestly, the visual result is already much better. Still a lot to improve, especially the walking animation, but the pipeline is getting closer: prompt → assets → prototype → test → iterate
Givros13,340 views • 3 months ago
I made my own version by giving Codex the reference images. First, I gave it the Japanese voxel game concept. Then, I gave it the asset sheet. Codex created the game assets based on that asset sheet, then used them to build the playable browser game. Comment “voxels japanese prompt” if you want the prompt.👇
Givros11,832 views • 3 months ago