Describe
Write what you want. "A red fox running", "an angry mushroom", "a wizard casting a spell". You can also attach a reference image.
Describe a voxel - the AI models it and animates it.
Type one sentence, get a fully animated 3D voxel model in roughly twelve seconds. Mesh, idle, walk - all generated. Add custom animations after, retouch the geometry voxel-by-voxel, or export straight to GLB or pixel art sprite sheet. No 3D modeling skill required.
Write what you want. "A red fox running", "an angry mushroom", "a wizard casting a spell". You can also attach a reference image.
The AI produces 4 mesh candidates + 4 animation candidates in parallel. ~12 seconds, 1 credit, 1 fixed-rate run.
You preview the 4 candidates side by side, animated in real time. Click the one you like, the others are discarded.
Retouch the silhouette voxel-by-voxel, add more animations (attack, jump, hit), or export to GLB / sprite sheet right now.
A single prompt produces a voxel mesh, idle and walk animations, ready to render. No prompt engineering tricks needed - short, plain English (or French) works.
Each step generates 4 candidates in parallel. You compare side by side, pick the best, the rest disappear. No regeneration loop, no luck-based reroll.
Drop a sketch, a reference photo, or a screenshot. A vision model describes it first, then the description is merged into your prompt - voxel style stays consistent.
1 credit for an entire guided run, regardless of how many AI calls happen internally. No surprise billing, no per-token pricing.
After the initial run, request any animation you want - attack, cast, jump, death - with the same 4-candidate flow. Animations stack in the same asset.
Animated GLB for 3D engines (Three.js, Unity, Godot, Babylon) or pixel art sprite sheet for 2D engines (Phaser, GameMaker, Aseprite).
The community gallery is fed by AI-generated models that users decided to publish - every entry is downloadable, browsable in 3D, and exportable. Good benchmark for what the prompts can produce.
A tool that turns a text prompt into an animated 3D voxel model. You type a sentence - "a fox running through grass", "a dragon breathing fire" - and the AI produces a 32×32×32 voxel mesh with idle and walk animations baked in. You can then retouch it in the editor, add more animations, or export it.
Yes - anonymous users get one free generation per day. A free account gets 2 credits to start. Each guided run (mesh + animations) costs 1 credit total, no matter how many internal AI calls happen behind the scenes.
A lightweight multimodal model handles retrieval and image-vision passes (cheap and fast), then a more capable model handles the heavy mesh and animation generation. Models are server-locked - there is no model picker - so cost and quality stay predictable.
Yes. Drop an image (a screenshot, a sketch, a reference) and the AI describes the main subject in 1-2 sentences before generating. You can also combine image + text - the description is merged into your prompt automatically.
Every guided run produces idle + walk by default. After the run, you can ask for any custom animation - "attack with a sword", "cast a spell", "death", "jump" - through a dedicated 4-candidate add-animation flow. All animations are stored in the same asset.
Yes. The output is exportable to animated GLB (glTF binary, ready for Three.js, Unity, Godot, Babylon, Blender) or to a pixel art sprite sheet PNG (8 directions, isometric, top, bottom - every animation × every frame).
Turn the AI output into a pixel art sprite sheet (8 directions, iso, top).
Export to glTF binary for Three.js, Unity, Godot, Blender.
Retouch the AI output voxel by voxel before exporting.
Already have a 3D model? Convert GLB/OBJ/FBX to voxel and animate it.
No signup needed for a first try. One sentence, twelve seconds.