The in-app image editing in DALL-E 3 is more useful than the generation itself for my workflow

The in-app editing with a selection brush to modify specific areas using text prompts converts it from a generation tool into a production tool. Generating something close to right and then precisely modifying the elements that are not right is faster than trying to prompt perfect generations on the first attempt.

Text generation within images is genuinely functional for greeting cards and signage concepts, with the caveat that occasional typos require fixing via the edit tool or regeneration. For anything where text accuracy is critical, regenerating is usually faster than editing.

The upload and edit personal photos capability, demonstrated by replacing a portrait background with a generated scene, is the practical photography use case that changes what is achievable without a full Photoshop workflow.

The on-screen message flagging DALL-E 3 as a legacy model with a new version rolling out is worth noting for anyone building workflows around specific model outputs.

Is inpainting part of your regular production process or are you still starting fresh with new generations each time?