Glossary
AI Video Glossary: Every Term You’ll See on GoCrazyAI.com
A plain-English dictionary of the models, modes, and concepts behind AI video, image, voice, and music — from text-to-video and voice cloning to Sora 2, Veo 3.1, and the AI Video Editor. Every term links to the page where you actually use it.
#
- 1080p1080p (1920×1080, Full HD) is the standard high-definition output most AI video models target — sharp enough for social, web, and most screens.
- 4K resolutionVideo or image output at roughly 3840×2160 pixels — four times the pixel count of 1080p.
A
- Accent & multilingualAI voices can speak with a chosen accent — American, British, Spanish, Indian, and many more — and in multiple languages.
- AI Ad Video GeneratorA tool that turns a product photo or a short prompt into a ready-to-run video ad — hook, motion, and pacing handled for you, output in vertical and square formats for social.
- Anime PFP GeneratorA tool that turns a prompt or a selfie into an anime profile picture in styles like cute, cool, aesthetic, kawaii, dark, cyberpunk, and manga.
- Anime ProGoCrazyAI's branded anime image models.
- Aspect ratioThe shape of a video or image expressed as width:height.
- Google AI OverviewThe AI-generated answer block Google now shows above traditional search results.
B
- BPMBeats per minute — the tempo of a track.
C
- Character CardA reference sheet that captures a single character from multiple angles and expressions so it stays consistent across images and videos.
- Character consistencyKeeping the same character — face, outfit, proportions — recognizable across multiple images or shots.
- Cinematic promptA prompt that explicitly directs the model to produce film-quality framing, lighting, and motion — using terms like 'cinematic 16:9', 'shallow depth of field', 'anamorphic lens', 'golden hour'.
- Creative 2.0GoCrazyAI's high-quality text-to-image model for generating images from a prompt.
- Credit-based pricingA pricing model where you buy a pool of credits and each AI generation deducts a fixed amount based on the model and length.
D
- Diffusion modelThe AI architecture behind most image and video generators.
E
- ElevenLabsElevenLabs is a voice-AI company whose models power GoCrazyAI's AI Voices and AI Song Generator.
F
- FLUX.2A text-to-image model available in GoCrazyAI's AI Image Studio, known for sharp detail, strong prompt adherence, and reliable text rendering inside images.
- Frame rate (FPS)How many still frames play per second of video, measured in FPS.
G
- Grok Imagine VideoxAI's Grok Imagine video model, available in both text-to-video and image-to-video modes.
H
- Hailuo (MiniMax)Hailuo is MiniMax's AI video model, known for smooth motion and strong prompt following on short clips.
- Happy HorseA fast video model offered on GoCrazyAI in three modes: image-to-video, text-to-video, and reference-to-video (using up to three reference images).
I
- AI Image StudioGoCrazyAI's all-in-one image workspace: generate, edit, upscale, restyle, build character cards, swap backgrounds, change camera angle, and apply effects.
- Image upscalingIncreasing the resolution of an image while reconstructing detail using an AI model — typically going from 1080p or 2K up to 4K or 8K.
- Image-to-videoA generative AI workflow where you upload a still image and an optional motion prompt, and the model animates the image into a short clip while keeping the subject on-model.
- InstrumentalA music track with no vocals — just the instrumentation.
K
- Kling 2.5 Turbo ProKuaishou's text-to-video and image-to-video model, in its 2.5 Turbo Pro variant.
- Kling 2.6 ProKuaishou's Kling 2.6 Pro video model, known for cinematic visuals and native audio that's generated to match the scene.
- Kling Omni Video O3Kling's Omni Video O3 model, which supports multi-shot generation with audio and lets you set both a start frame and an end frame so the clip lands on a target image.
L
- Luma Dream MachineLuma's Dream Machine is an AI video model recognized for fluid camera motion and realistic physics in short clips.
M
- Motion ControlA feature that lets you direct the camera in an AI-generated clip — pans, orbits, push-ins, and zooms — instead of leaving movement to chance.
- Multi-referenceSupplying more than one reference image to guide a generation — for example a character, a setting, and a prop — so the model honors several visual anchors at once.
N
- Google Nano BananaGoogle's text-to-image model used inside GoCrazyAI's AI Image Generator.
- Native audioAudio that a video model generates together with the picture, so sound effects, ambience, or speech match the visual without a separate dubbing step.
- Negative promptA list of things you want a generative model to avoid — 'no text, no extra fingers, no blur'.
P
- AI PodcastA studio that generates multi-voice AI podcasts from a topic or script — distinct AI voices hold a natural-sounding conversation, output as a single mixed audio file up to about an hour long.
- People Also AskGoogle's expandable Q&A panel that appears in search results, surfacing related questions other users typed.
- PikaPika is an AI video generator known for fast, stylized short clips and playful effects.
- Plush AIAn AI image style that renders any subject as a soft, stuffed-toy plushie — rounded felt textures, stitched seams, button eyes.
- Pro Video StudioGoCrazyAI's advanced video workspace, where you pick a model, set duration, resolution, and aspect ratio, supply start/end frames or multiple references, and use power features like native audio and video extend.
- Prompt engineeringThe practice of phrasing inputs to generative AI models so they produce useful, on-target outputs.
R
- AI relightingChanging the lighting of an existing photo using an AI model — switching daytime to golden hour, flat lighting to studio, or adding neon.
- Reference imageA still image you provide to a generative model as a visual anchor.
- RunwayRunway is an AI video platform whose Gen-series models (Gen-3, Gen-4) are widely used for text-to-video and image-to-video.
S
- AI Song GeneratorA tool that generates full, original songs and instrumentals from a text prompt — pick a genre, mood, and tempo, with optional AI-written lyrics or your own cloned voice on vocals.
- OpenAI SoraOpenAI's flagship text-to-video model.
- SeedA number that fixes the random initialization of a generative model.
- Seedance 1 ProByteDance's professional cinematic AI video generation model.
- Seedance 2.0The second-generation Seedance video model from ByteDance, supporting both text-to-video and image-to-video with native audio at up to 1080p.
- Sora 2 ProThe higher-fidelity tier of OpenAI's Sora 2 text-to-video model.
- Start & end framesImages you supply to fix how a generated clip begins and ends.
- SubtitlesOn-screen text of the spoken audio, either auto-generated and burned into the video or kept as a separate track.
- SynthesiaSynthesia is an AI avatar-video platform aimed at corporate training and presenter-style videos, where a digital avatar reads a script.
T
- AI talking avatarA still photo turned into a video where the person or character appears to speak, with mouth movements, expressions, and head motion synced to a voice.
- Text-to-imageA generative AI workflow where you describe an image in plain English and the model renders it as a still picture.
- Text-to-speechSynthesizing spoken audio from written text using an AI voice model.
- Text-to-videoA generative AI workflow where you describe a scene in plain English and the model renders it as a short video clip.
V
- AI Video EditorGoCrazyAI's editor for finishing AI-generated video: add voiceover, music, and sound effects, auto-generate and burn in subtitles, and layer text or overlays — then export a single ready-to-publish file without leaving the platform..
- AI VoiceGoCrazyAI's voice library — 160+ premium AI voices for narration and dubbing, plus voice cloning from a short sample and voice design from a text description.
- Google VeoGoogle DeepMind's flagship text-to-video model.
- Vertical videoVideo shot or rendered in 9:16 aspect ratio — taller than it is wide.
- Video extendA feature that continues an existing clip past its original length, generating new frames that flow on from the last ones.
- Voice cloningA generative AI process that builds a digital model of a real voice from a short audio sample (typically 30–60 seconds), then synthesizes new speech in that voice from text.
- Voice designCreating a brand-new synthetic voice from a text description — age, tone, accent, energy — rather than cloning a real one.
- Voice IsolatorA tool that separates a clean voice track from background noise or music in an audio file.
- VoiceoverA spoken narration track laid over video.
W
- WanAlibaba's Wan family of video models (versions 2.5, 2.6, and 2.7).
- Watermark-freeOutput with no overlaid logo or watermark, so the file is ready to publish or use commercially as-is.