Adobe Certified Professional - Adobe Firefly key terms

A category of artificial intelligence that creates new text, images, audio, or video by learning patterns from large training datasets rather than only classifying or predicting existing data.

Adobe's family of generative models that produce images, vectors, video, and audio, trained on licensed and public domain content and designed for commercial-safe output.

A generative technique that learns to remove noise step by step, then reverses the process to build a coherent image starting from random noise guided by a text prompt.

The written instruction describing the subject, style, lighting, composition, and mood you want a generative model to produce.

Text that tells the model what to avoid or exclude from the result, such as unwanted objects, artifacts, or styles.

The practice of emphasizing or de-emphasizing parts of a prompt so certain words influence the output more strongly than others.

The proportional relationship between an image's width and height, such as 16:9 or 1:1, which determines the shape of generated output.

A starting value that initializes the random noise in a generation, allowing a user to reproduce or make small variations of a prior result.

A new output generated to resemble an existing result while differing in details, often used to explore alternatives from one prompt.

An input image or preset used to guide the visual look of a generation, such as color palette, texture, or artistic treatment, without copying its subject.

An input image used to control the layout or composition of a generation, so the new output follows the same arrangement of shapes and elements.

A setting that tells Firefly which kind of output to produce, such as photo, art, or graphic, adjusting the model's rendering approach.

A Firefly feature that inserts newly generated content into a selected region of an image using a text prompt, matching surrounding context.

The process of filling or replacing a masked area of an image with generated content while keeping the rest of the image unchanged.

Extending an image beyond its original borders with generated content, often called generative expand, to change framing or aspect ratio.

A Firefly capability that enlarges a canvas and generates new surroundings that blend with the existing image.

The defined area of an image where generative edits are allowed to occur, protecting everything outside the selection from change.

A tool that erases unwanted objects or people from a photo and reconstructs the background behind them using generated content.

Increasing an image's pixel dimensions while using generative or interpolation methods to add detail and reduce blur.

A Firefly feature that applies generated textures, materials, and styles to typed text while preserving letterform legibility.

A Firefly feature that applies new color palettes to vector artwork while keeping the original shapes and structure intact.

Artwork built from mathematical paths rather than pixels, so it can be scaled without loss of quality and edited by shape.

An image made of a grid of pixels, where quality depends on resolution and enlarging too far reveals visible pixelation.

The amount of detail an image holds, usually expressed as pixel dimensions or pixels per inch, affecting print and screen quality.

Firefly capability that creates short video clips from a text description or a starting image, producing motion over time.

A video generation mode that animates a supplied still image, using it as the first frame and inferring subsequent motion.

A sequence of sketched or generated frames that plans shot order, camera movement, and timing before full production.

Firefly features that synthesize speech, sound effects, or background music from text descriptions or prompts.

Technology that converts written text into spoken audio, often with selectable voices and adjustable pacing.

A prebuilt layout or design starting point that can be customized with your own content, used in design workflows for speed and consistency.

Tamper-evident metadata attached to digital content that records its origin, edits made, and tools used, often called a nutrition label for content.

The Coalition for Content Provenance and Authenticity, an industry standard that defines how origin and edit information is attached to media files.

The recorded history of where content came from and how it was modified, used to verify authenticity and build trust.

Using generated output in business, advertising, or monetized work; Firefly models are trained so outputs are generally intended to be safe for this.

The legal terms governing how generated or stock content may be used, including restrictions on redistribution and resale.

The collection of images, text, or media a model learns from; Firefly is trained on licensed and public domain content to reduce legal risk.

Systematic skew in model outputs that overrepresents or underrepresents groups, often reflecting imbalances in training data.

Principles and practices for developing and using AI safely and fairly, including transparency, accountability, and avoiding harmful or deceptive output.

Synthetic or manipulated media that convincingly depicts real people saying or doing things they did not, often used to mislead.