Meitu Image Generation and Editing Alternative

Problem

E-commerce designers, retouchers, and operators often need to turn product photos, reference images, or short selling-point descriptions into publishable main images, detail pages, posters, or social-media covers. Writing API calls, uploading assets, polling tasks, and downloading results interrupts the visual iteration loop. This skill packages prompt, reference images, and model parameters into one command for production tasks with a clear asset purpose.

How it works

It runs on AI Hive OpenAPI and calls the fixed model public_model_nano_banana_pro. It supports two input modes:
- Text-to-image: use --prompt to specify purpose, subject, composition, style, lighting, required text, and output specs.
- Reference-image editing: use --image to provide product, material, or composition references, and state each image's role to reduce subject drift.

The workflow includes initializing an API Key, submitting the task, polling taskId, and downloading the result. --batch creates multiple images, --param key=value passes real-time model parameters, and --routing selects COST_FIRST, SPEED_FIRST, or SUCCESS_FIRST. Output defaults to ~/Downloads/AiHive.

Boundaries

It fits clear deliverables such as e-commerce main images, detail pages, ad KV, and social covers, but is not ideal for open-ended artistic exploration or final assets requiring exact typography. Text that must appear verbatim should be quoted, and final output should be reviewed manually. Confirm real-time pricing before batch generation; if the backend model or parameters change, check the live imageConfig before submitting.