DALL-E 2 inpainting and outpainting are the features that turned it from a generation toy into something more useful
Text-to-image generation is the feature everyone knows about. The inpainting and outpainting capabilities are the ones that change how you use it for practical work.
Inpainting for editing specific parts of an image by removing a selected area and filling with AI-generated content based on a new prompt is targeted editing that separates it from basic generation. You are not regenerating the whole image to change one element.
Outpainting for expanding images beyond their original borders by generating new content to extend the scene is the aspect ratio and framing flexibility that solves a common photography problem. An image composed for one format can be extended to work in another without losing the original composition.
Image variations for generating different versions of an existing image are the iteration tool that helps when the composition is right but the specific style or element is not quite what you need.
The combination of these three capabilities alongside basic generation makes DALL-E 2 more than a prompt-to-image tool.
What capability do you use most in practice: generation from scratch, inpainting, outpainting or variations?