Sprite sheet generator for 2D pixel art characters and maps. Built with fal.ai.
| Walk Cycle | Jump Animation | Attack Animation |
|---|---|---|
- Side-Scroller generates walk, jump, attack, and idle sprite sheets, plus an optional 3 layer parallax background.
- Isometric (RPG) generates walk sheets for three directions (down, up, side), matching attack sheets for the same directions, an idle sheet, and a full top-down world map to explore.
- Character generation from a text prompt or by converting an uploaded image into pixel art.
- Sprite sheets rendered as 2x2 grids per animation, with consistent character identity across frames.
- Backgrounds generated to match your character. Side-scroller mode produces a 3 layer parallax scene (sky, midground, foreground). Isometric mode produces a single large top-down map.
- Background removal via Bria for sprite sheets and parallax midground and foreground layers.
- Frame extraction with adjustable grid dividers for each sprite sheet.
- Animation preview with adjustable FPS.
- Per sprite size sliders in the sandbox so you can correct scale without regenerating.
- Layer position sliders for side-scroller custom backgrounds, so you can nudge each layer vertically if alignment is slightly off.
- Map size slider for the isometric map, so you can rescale the world relative to the character.
- Per layer regeneration for the 3 parallax layers, so you can retry one without redoing the others.
Pick the model once at the top of Step 1 and it applies to every generation in the flow:
- Nano Banana Pro (
fal-ai/nano-banana-proand/edit) - GPT Image 2 (
openai/gpt-image-2and/edit)
- Install dependencies:
npm install- Create a
.env.localfile with your fal.ai API key:
FAL_KEY=your_api_key_here
Get your API key at https://fal.ai/dashboard/keys
- Run the development server:
npm run devD/→Walk rightA/←Walk leftSpaceStop
A/←Walk leftD/→Walk rightW/↑JumpJAttack
W/↑UpS/↓DownA/←LeftD/→RightJAttack
- Next.js 14
- React 18
- fal.ai (Nano Banana Pro, GPT Image 2, Bria background removal)
- HTML Canvas