Term · Intermediate
Img2Img
Generating a new image guided by an input image plus a prompt.
Img2img feeds an existing image into a model as a starting point, so the output keeps its composition or structure while the prompt reshapes style and detail. A denoising strength controls how far it strays from the source. Creators use it to restyle photos, iterate on sketches, and lock composition while changing look, more control than starting from text alone.
Also called image to image, img-to-img, i2i
Related
- Text-to-ImageGenerating still images from a written prompt using AI.
- ControlNetAn add-on that steers image generation using structure like pose, depth, or edges.
- InpaintingMasking part of an image and letting AI redraw it, to remove, replace, or extend content.
- Stable DiffusionOpen-source image model you can run locally with deep customization and control.
- Latent SpaceThe compressed abstract space where a model represents concepts before rendering output.