5.2.2. Editing Generated and Existing Media
💡 First Principle: Editing is a constrained form of generation — you start from existing media and apply targeted changes rather than producing from a blank prompt. The platform exposes specific editing controls: inpainting (regenerate a region), mask-based edits (confine changes to a marked area), and prompt-driven modification (change the image according to an instruction), plus workflows to edit generated video.
Inpainting and mask-based editing let you change part of an image while preserving the rest — supply the source image, mark the region (the mask), and describe the replacement, and the model regenerates only the masked area. Prompt-driven modification applies an instruction to transform an image ("make the background a sunset") without an explicit mask. Video editing workflows extend the same idea to generated video segments. The skill is matching the requirement to the right control: a localized change ("replace just the logo") points to mask-based inpainting; a global restyle points to prompt-driven modification. Selecting and applying the appropriate generation and editing controls the platform provides is itself a named exam objective.
⚠️ Exam Trap: A requirement to change only a specific region of an image is a mask-based / inpainting task, not a full regeneration — regenerating the whole image from a new prompt loses the parts that were supposed to stay. Conversely, a whole-image restyle does not need a mask. Matching the scope of the change (local versus global) to the editing control (masked versus prompt-driven) is the decision.
Reflection Question: A team must replace the product label in a generated photo while keeping everything else identical. Why is mask-based inpainting the correct control rather than re-running the original generation prompt with the new label described?
