sprite-gen: Generate Game Sprites and Animation Atlases

Overview

sprite-gen is a Python CLI and Codex/Claude skill for turning one character image into game-ready sprite assets. Its atlas pipeline generates state rows, cleans up transparency, extracts frames, and builds an atlas with a machine-readable layout manifest. For animated loops, it uses Grok Imagine to create video clips and then cuts them into transparent animations.

The tool targets game developers and coding-agent users who need more than a generated sprite sheet: it includes frame extraction, optional visual curation, runtime-oriented exports, and tools for processing existing assets. Generation uses provider credentials supplied by the user, while some workflows also rely on external command-line tools.

Key Features

• Generate sprite states from a base image, then extract transparent frames and compose a runtime atlas with frame-layout metadata.
• Create per-state motion loops from still images using Grok Imagine video generation and frame-processing commands.
• Review, reject, or adjust generated frames in an optional curation webview before composing an atlas.
• Apply deterministic palette-based recoloring and export atlases for formats including Aseprite, Phaser, and Flame.
• Process existing art with utilities for cutouts, sheet slicing, atlas unpacking, repeating backgrounds, and projected shadows.
• Inspect animation timing and contact evidence, and optionally compose existing assets into rendered scenes.

Use Cases

• A solo game developer wants to turn one character illustration into an atlas with explicit frame rectangles for a game engine.
• A developer using Codex or Claude wants an agent-accessible workflow to generate, inspect, and prepare sprite assets.
• An artist needs transparent idle, attack, or other motion loops from a still image and is prepared to use the video pipeline.
• A team with existing sprite sheets wants to slice, recolor, export, or inspect assets without regenerating them.

Getting Started

python3 -m venv .venv && source .venv/bin/activate
pip install -e .
sprite-gen --help

See the README for pipeline commands, provider setup, and workflow documentation.

Considerations

The project is less than six months old and is changing quickly, with recent commits and frequent releases. Although it has several contributors, nearly all measured contributions come from one person, so maintenance continuity is a consideration. The measured health data reports tests but no CI. The atlas workflow needs Python 3.11 or newer; video workflows additionally document requirements such as ffmpeg, img2webp, and a Grok login or API key. Generated assets rely on user-provided credentials, and API-key calls may incur per-call costs. The README also cautions that facing detection can be wrong and that cyclic locomotion remains experimental unless motion QA passes.