Text-to-image or image editing?
Choose by whether you are inventing a new visual direction or starting from an existing image that matters to the brief.
Use text-to-image for a new concept
A written prompt describes the scene, subject, composition and style. This is useful for exploring a visual idea before an exact source image exists. A prompt containing a product name does not prove that the output will preserve that product’s real shape, markings or specifications.
Use an editing workflow for source-based changes
An editing route accepts the source input specified in its workspace. Check whether it expects one image, multiple references or a mask and whether the requested change is supported. Describe what should change and what should remain. Keep the source available for comparison rather than relying on memory.
Review differently for each job
For a concept, review composition and fit with the brief. For an edit, also compare identity, layout, labels and untouched areas with the source. Neither route removes the need to check facts or permissions. Regenerate or use the original if a required feature changes.
Choose the actual route
The directory separates many image tasks and variants. Inspect the required inputs in the selected workspace, then review the credit estimate after sign-in. The published MiniMax example illustrates a text-to-image preset; it is not an editing benchmark.
| Question | Text to image | Image editing |
|---|---|---|
| What do I start with? | A written scene description | The source image(s) required by the route |
| Main decision | Explore a new visual concept | Change an existing visual using supported inputs |
| Important review | Composition and brief following | Source fidelity and intended changes |
| Exact product preservation guaranteed? | No | No; inspect the actual result |