AI Image Editing Glossary

Plain-language definitions for the AI image editing terms you will run into - from face swap to upscaling.

Key takeaways

  • Know the difference between a face swap and a deepfake.
  • Upscaling changes size; enhancement changes quality.
  • Transparency, alpha channel and inpainting come up constantly in editing.
  • A seed plus identical settings reproduces the same result, which is handy for fine-tuning an edit.

Face swap & deepfake

Face swap: replacing a face in one photo with a face from another, with consent, as a creative edit - try it on AI Face Swap. Deepfake: synthetic media designed to deceive, usually without consent. Same family of tech, very different intent - see our comparison.

Upscaling & enhancement

Upscaling increases an image's resolution by reconstructing detail (Image Upscaler). Enhancement improves quality - light, colour, sharpness - at the same size (Image Enhancer). They solve different problems.

Restoration, colourisation & background removal

Restoration repairs old or damaged photos. Colourisation adds natural colour to black-and-white images (Photo Colorizer). Background removal cuts the subject out onto transparency (Background Remover).

Formats & technical terms

Alpha channel: the transparency layer in PNG and WebP. Inpainting: filling in or removing parts of an image. Prompt: the text description used to generate an image (AI Image Generator). Aspect ratio: the width-to-height proportion of an image.

Generation and model terms

  • Diffusion model - the type of AI behind most modern image tools; it starts from random noise and gradually refines it into a coherent image.
  • Denoise - the step-by-step removal of that noise; the denoise strength controls how much an input image is changed.
  • Latent space - a compressed internal representation of an image the model works in, rather than on raw pixels directly.
  • Seed - a number that fixes the starting randomness, so reusing the same seed and settings reproduces the same result.
  • Prompt - the text description guiding generation on the AI Image Generator.
  • Super-resolution - the technical name for AI upscaling that reconstructs new detail rather than stretching pixels.

Editing and image-property terms

  • Inpainting - filling a selected area with new, context-aware content to remove an object or repair damage.
  • Outpainting - extending an image beyond its original borders, inventing plausible surroundings.
  • Mask - a black-and-white map marking which pixels an edit should affect and which to leave untouched.
  • Aspect ratio - the width-to-height proportion of an image, such as 1:1 square or 16:9 widescreen.
  • DPI - dots per inch; the pixel density that determines how large a photo can print sharply.
  • Lossy vs lossless - JPEG is lossy and discards data to save space, while PNG is lossless and preserves every pixel, useful for cut-outs from the Background Remover.

Frequently Asked Questions

Is a face swap a deepfake?
It uses related technology, but a consented creative swap is not a deceptive deepfake.
What's the difference between upscaling and enhancing?
Upscaling changes size by adding detail; enhancement improves quality at the same size.
What is an alpha channel?
The transparency information in formats like PNG and WebP.
What is a seed in AI image tools?
A seed is the number that fixes the random starting point of generation, so reusing the same seed with identical settings reproduces the same image.

Keep going

More guides