← Generative & Multimodal AI
07 · Generative & Multimodal AI · Topic
What is Image Generation?
Image generation uses AI models to create or transform visual content from prompts or other inputs.
What is it?
Image generation is the use of an AI system to create a new image or edit an existing image from supplied conditions.
Possible inputs include text, an image, a mask, a layout, or other conditions. The output may be a new image or an edited image.
Think of it like...
Giving a visual artist a description and asking them to create an image, except the AI produces pixels through learned mathematical patterns.
How it works
1 · Prompt / Condition
Provide text, an image, a mask, a layout, or another condition.
2 · Encode Input
Turn the input into a representation the model can use.
3 · Generative Model
Apply patterns learned from visual data.
4 · Build / Refine Visual Representation
Construct or refine a result that matches the conditions.
5 · Decode Image
Convert the representation into pixels.
6 · Output
Return a new image or an edited image.
Real-world examples
Everyday example · Image editing
Describe an unwanted object, and the system fills that area with a consistent replacement.
Product example · Mockup creation
A team describes a product scene and generates concept images for early design exploration.
What it is NOT
Image Generation ≠ Image Search
Search finds existing images; generation creates new visual content.
Image Generation ≠ Computer Vision
Computer vision interprets images; image generation produces or transforms them.
Image Generation ≠ Diffusion Model
Image generation is a capability; diffusion is one model family used by some systems.
Related concepts
Image Generation
├── Generative AI
├── Multimodal AI
├── Diffusion Models
└── Computer Vision
Remember this
Image generation creates new visual content from learned patterns and supplied conditions.