The outcome

Produce a controlled image edit whose source, mask, instructions, transformations, and approvals remain auditable.

Stability AI is a generative image, editing, upscaling, control, video, audio, and 3D model API platform. Building applications that generate, edit, upscale, control, and transform images through Stable Image services and related Stability AI models. This guide narrows that broad capability into one repeatable outcome, with checkpoints that keep the source material and your judgment in the loop.

Before you begin

Set the boundary before the tool starts.

Choose one real task, identify who will use the result, and decide what evidence or test will make the result acceptable. Gather only the source material needed for that task. If the work contains confidential, personal, regulated, or client-owned information, confirm that the platform and account are approved before sharing it.

Troiana principle

AI should make the work easier to inspect. If the workflow removes the source, the owner, or the review step, redesign the workflow.

Step by step

A workflow you can repeat.

  1. 01

    Confirm ownership or permission for the image and depicted people, then define the exact edit, protected regions, output size, and prohibited changes.

  2. 02

    Remove unnecessary metadata, retain an immutable original, and create a precise mask or prompt with supported dimensions and request size.

  3. 03

    Call the narrowest edit endpoint with a server-side key, explicit output format and seed, and no unrelated reference images or sensitive data.

  4. 04

    Compare output against the original for unintended identity, logo, text, background, lighting, and safety changes at full resolution.

  5. 05

    Store the original, mask, prompt, endpoint, seed, and approved result separately; disclose material edits and delete rejected assets on schedule.

Working standard

What good use looks like.

  • Keep an immutable original.
  • Use the narrowest edit endpoint.
  • Document material alterations.

Generated and edited media can contain unsafe, biased, deceptive, infringing, or unwanted likeness content. Use current endpoints, protect keys, confirm rights and consent for inputs, retain originals, minimize uploaded data, record model settings and seeds, review outputs at full resolution, disclose material edits, handle moderation and rate limits explicitly, and track model deprecations, licensing, and credit use.

Official references

Check the current product documentation.

Features, plan limits, availability, and data controls change. These official pages are the starting points used for this collection.