data-slots=image, heading, text
data-background=rgb(80, 190, 233)
Adobe Firefly API - Firefly Services
This is the user documentation for Adobe Firefly API. Firefly is the natural extension of the technology Adobe has produced over the past 40 years, driven by the belief that people should be empowered to bring their ideas into the world precisely as they imagine them. Describe your creative vision in your own words and we will bring it to life, saving you time and effort.
With Firefly, you can ideate, create, and communicate while significantly improving creative workflows using generative AI. You might have already tried out the Firefly web app, along with features powered by Firefly in Adobe’s flagship apps and Adobe Stock. And now it is available to you as an API. Integrate them with your workflow as it suits your needs.
data-slots=link, text
data-slots=link, text
data-slots=link, text
Custom Models API
Generate brand-aligned image variations at scale with Firefly's Custom Models API. Train models on your brand's aesthetics, characters, or products and use them in your API requests to keep visuals consistent across campaigns and channels.
What is the Custom Models API?
The Custom Models API lets you:
- Train subject or style models — Capture specific characters, products, or objects (subject models) or aesthetic qualities like color palettes, patterns, and illustrative style (style models) so generated images match your brand.
- Generate images with a custom model — Call the Generate Image API with a custom model asset ID so outputs reflect your model's learned characteristics.
- Use style presets and controls — Combine custom models with style presets and image size parameters for fine-tuned, on-brand results.
- Manage and share models — Each custom model has a unique asset ID for versioning and reuse; you can grant applications access so they can list and use your models via the API.
Why choose Custom Models?
- Scale brand-consistent visuals — Replicate distinctive brand look and feel across many images without manual art direction.
- Consistent characters and products — Keep mascots, products, or key objects recognizable in any scenario you prompt.
- Secure hosting and reuse — Models are hosted by Adobe with clear versioning; reference them by asset ID in your workflows and share them with your technical accounts as needed.
To get started, see the Custom Models overview, the Custom Models Generate Image API tutorial, and Grant apps access to a Custom Model.
What's new with Image Model 5?
The Image5 model is Firefly's latest model and is a leap forward in realism, lighting, and composition with native 4 MP resolution for sharper, more expressive results across photography, text, and illustration.
It also supports the new Instruct Edit capability, which enables users to generate and refine in one seamless workflow using natural language to make precise and intuitive edits.
If you've already implemented image generation with the Firefly API, you can migrate to the Image5 model by following the Migrating to Image5 guide.
Composite Operations APIs
Blend product shots and objects into generated scenes with complementing tones, colors, lighting, shadows, and textures.
This service offers the Object Composite API, Adaptive Composite, and Precise Composite.
Composites with Adobe Firefly allows you to upload an image, such as a product photo, and use a text prompt to generate a composited scene for the product.
The Composite Operations APIs can be used to:
- Create captivating product shots with different setups, backgrounds, and visual design variations.
- Generate engaging social media creatives with custom images and product photos.
- Help customers visualize scenarios by showcasing the product in different settings and environments.
What are the Composite Operations APIs?
The Precise Composite API is improved over previous versions of the service in the following ways:
-
Higher Fidelity and Realism - The new ME model significantly outperforms the previous MD model in object-edge accuracy, lighting realism, and contextual scene generation.
-
Enhanced Masking - Enhanced masking (leveraging the latest Photoshop mask models) eliminates the "halo" effect common in v3 outputs.
-
True Visual Integration via Harmonization - Generative Harmonization reintroduces and enhances lighting and shadow correction for composite realism. Developers can control harmonization strength (0–100%) or toggle the feature off entirely to fine-tune visual output.
Use Adaptive Composite to:
-
Bring-Your-Own-Background to Adaptive Composite - Using Adaptive Composite, customers can now composite products into existing or custom-generated backgrounds—such as brand-specific environments produced with Firefly Custom Models. This allows greater creative control, brand consistency, and multi-scene reuse across campaigns.
-
Intelligent Object Rotation - The Adaptive Composite model analyzes the target background and automatically adjusts the object's rotation to match environmental cues (for example, perspective or lighting direction).
To get started, see the Object Composite API Feature Guide.
Upscale API (beta)
Enhance image resolution and quality quickly and effortlessly, without pixelation or retakes.
Demand for high-resolution images is growing across digital platforms, but traditional upscaling often causes pixelation and quality loss. The Upscale API delivers automated image enhancement that saves time and resources. Integrate it into Adobe products or use it in your own workflows to improve image quality where it matters, while keeping details crisp and accurate.
What can the Upscale API do?
The Upscale API is designed for workflows that feature these use-cases:
User Assets Upsampler
Increase the resolution of your own photos and artwork while preserving detail and avoiding artificial or hallucinated content. Use this use-case to:
- Preserve the original: Follow the source image closely and avoid introducing new or artificial elements.
- Transform low-resolution sources: Turn small or low-quality images into high-resolution versions suitable for print, display, or downstream workflows.
- Support multiple scale factors: Upsample to 2×, 4×, or 6× depending on your needs.
Generated Image Upsampler
Maintain crispness and correct fine details in images created by generative AI (e.g., textures, patterns) when zooming in.This capability focuses on two dimensions of quality:
- Crispness: The upscaled image retains the sharpness of the original generation.
- Correctness: Fine details (e.g., textures like flowers, grass, or fabric) stay coherent when you zoom in, instead of breaking down into a messy or blurry appearance.
To get started, see Getting Started for authentication and setup, and the API Reference for Upscale endpoints and schemas.
Discover
data-slots=link, text
data-slots=link, text