FAQ

What models are available?

spritesynth offers three AI models:

• Pixflux — Fast generation, good quality. Best for rapid iteration and prototyping.
• Pixen — Slower but higher quality. Best for final production assets.
• Bitforge — Balanced speed and quality. A solid all-around choice.

You can switch models per generation or set a default in your account settings.

Do I need a GPU?

No. All inference runs on cloud GPUs via RunPod serverless endpoints. You generate sprites from a browser or API — the heavy compute is handled remotely. No local GPU required.

Can I use generated art commercially?

Yes. All sprites generated on spritesynth are yours to use in commercial projects — games, apps, merchandise, or any other product. No attribution required.

What file formats are supported?

Export formats:

• PNG (single frame)
• Sprite sheets (horizontal, grid)
• GIF (animated)
• Unity sprite sheets + metadata JSON
• Godot sprite sheets + .tres resource files
• Aseprite .aseprite JSON format
• Tiled tileset JSON

Import formats:

• PNG, JPG, WebP for style references and init images

How do credits work?

Every generation costs credits based on the model and output size:

The free tier gives you 40 fast generations that never expire, plus 5 slower generations each day (resets daily, does not carry over). Paid tiers include priority queue access.

Can I self-host?

Yes. spritesynth supports self-hosting with Docker and RunPod. You'll need:

• A PostgreSQL database
• Redis for caching and queues
• An S3-compatible object store
• RunPod account for GPU inference

See the Self-Hosting section for deployment instructions.