Plain-English definitions of the technical terms that come up when you read AI image documentation, blog posts and prompt guides.
A LoRA is a small add-on file (usually 50-200 MB) that teaches a base AI image model a specific style, character or visual concept without retraining the whole model. The base mode…
Read full definition →ControlNet is an add-on architecture that lets you constrain an AI image generation to a specific pose, depth map, edge map or composition. Instead of relying on the prompt alone, …
Read full definition →A diffusion model generates an image by starting from pure noise and gradually denoising it over many small steps until a coherent image emerges. The model learns this denoising pr…
Read full definition →CFG Scale controls how strictly an AI image model follows your prompt versus how much creative freedom it takes. Low values (1-4) produce loose, dreamlike results that may ignore p…
Read full definition →A negative prompt is a separate text field where you list things the model should NOT include in the image. Useful for avoiding watermarks, extra fingers, blurry artifacts and unwa…
Read full definition →Inpainting lets you mask a specific area of an image and have the AI regenerate only that area while keeping the rest untouched. The classic use cases are removing an unwanted obje…
Read full definition →Image-to-Image (often shortened to img2img) starts the diffusion process from an existing image instead of pure noise. The model preserves the rough composition and colors of the i…
Read full definition →A seed is the random number that determines the initial noise pattern a diffusion model starts denoising. Using the same seed with the same prompt and the same model produces an al…
Read full definition →Compare 14 official AI image generators — new models added monthly.