AI media glossary

The vocabulary of AI media, in plain terms. Each entry links to the guide that goes deeper.

A–C

Allowance — the credits included with your plan that refresh each cycle; spent before anything else. See Credits and plans.

Aspect ratio — the shape of the frame (16:9, 9:16, 1:1). Picked per output; see Output formats and limits.

Batch — generating several images at once to pick the best. See Output formats and limits.

Credits — what you spend per generation; cost depends on model, length, and resolution. See Credits and plans.

D–L

Denoise — cleaning background noise from an audio source before re-voicing it. See Music, sound effects & dialogue.

Dubbing — translating and re-voicing existing audio or video into another language. See AI dubbing.

Effect — a one-tap template that turns a photo into a themed clip. See One-tap Effects.

Gallery — where your generated results collect, ready to reuse, edit, or carry into another tool. See Image generation basics.

Guest — browsing before you sign in; you can explore but not generate. See Account and sign-in.

Image-to-video — animating a still image you provide. See Image-to-video guide.

Inpainting (annotate-edit) — regenerating just a marked region of an image. See Annotate edits & reference images.

Instrumental — generated music with no vocals, so it sits under a voiceover. See Music, sound effects & dialogue.

Lip-sync — matching a character's mouth to an audio track. See Talking avatars & lip-sync.

M–R

Model — the engine that does the generating; each has different strengths. See Choosing a model.

Motion brush — painting the area of a shot where motion should happen, for finer control. See Kling 3.0 motion brush.

Motion control — driving a still character with motion from a reference video. See Text-to-video basics.

Multi-shot — several camera cuts produced in one video generation. See Kling 3.0 multi-shot and Prompting for video.

Native audio — sound generated together with the picture by an audio-capable video model, like Veo 3.1 native audio. See Choosing a model.

Negative space — empty area left in an image, often for placing text later. See Prompting for images.

Prompt — the text description that steers a generation. See Prompting for images and Prompting for video.

Queue — a short wait when load is high; Pro tiers get priority. See Troubleshooting generations.

Reference image — an attached image that guides style, subject, or composition; mention one with @. See Annotate edits & reference images.

Reference-to-video — carrying a subject or style from reference media into a generated clip. See Text-to-video basics.

Resolution — the pixel size of the output (e.g. 1K–4K for images); model-dependent. See Output formats and limits.

Reward credits — bonus credits that can expire; spent after your allowance but before top-ups. See Credits and plans.

S–Z

Start / end frame — the first (and optionally last) image of a video; the model animates from it. See Image-to-video guide.

Text-to-image — generating an image from a written description. See Image generation basics.

Text-to-video — generating a clip from a written description. See What is text-to-video AI?.

Upscaling — increasing the resolution and detail of an existing image or video. See Editing & cleanup tools.

Voice changer — re-voicing a recording in a different voice while keeping the original timing. See Music, sound effects & dialogue.

Voiceover (text-to-speech) — generating spoken audio from a script. See Voiceover and audio basics.

Can't find a term? The FAQ covers the common questions.