Glossary
Plain-English definitions for AI video, image, and generative model terms. Updated as the field moves.
Text-to-video is the task of generating a video clip directly from a text prompt, using a generative AI model that maps language to motion, scene composition, and visual style.
Image-to-video is the task of animating a single still image into a short video clip, with motion generated by AI based on optional text prompts that describe the desired action or camera move.
Face swap is an AI image and video editing technique that replaces the face of a person in a source media with the face of a target person, while preserving the original expressions, lighting, and pose.
AI lip sync is the task of automatically generating mouth and jaw movements in a video that match a given audio track, so a person's lips appear to speak the audio naturally.
AI upscaling is the task of increasing the resolution of an image or video while reconstructing plausible detail, using a neural network trained to predict what high-resolution content would look like.
Inpainting is the task of filling a masked region of an image with new AI-generated content that blends seamlessly with the surrounding pixels — used for object removal, background extension, and content replacement.
LoRA (Low-Rank Adaptation) training is a parameter-efficient fine-tuning technique that adds small trainable matrices to a frozen base model, letting you teach an AI generator a new subject, style, or character without retraining the whole model.
A diffusion model is a type of generative AI that creates images or video by progressively denoising random noise, guided at each step by a learned prediction of what the final image should look like.
Prompt engineering is the practice of crafting input text for AI models to consistently produce desired outputs — covering structure, specificity, style references, negative prompts, and parameter control.
Flux Pro is the highest-quality variant of the Flux image generation model family from Black Forest Labs (founded by alumni of the original Stable Diffusion team), released in 2024 and updated through 2026.
Nano Banana is the codename for Google's Gemini 2.5 Flash Image model — an AI system that generates and edits images from natural-language instructions, with conversational multi-image composition as its standout capability.
Sora 2 is OpenAI's flagship text-to-video generation model, with Sora 2 Pro the highest-quality variant, capable of producing cinema-quality clips up to 20 seconds long from a text prompt.
Veo 3 is Google DeepMind's flagship text-to-video model, producing high-fidelity cinematic video with native audio generation — a capability that distinguishes it from most competing video models.
Kling AI is a family of text-to-video and image-to-video generation models developed by Kuaishou, the Chinese short-video company, with Kling 2.6 the current flagship as of 2026.