New ways to use the Midjourney Image Editor: upload and edit images, erase/restore, and expand the canvas

The Midjourney Image Editor has finally opened up the “upload an image and then edit it” workflow, no longer limited to editing only images you generated yourself. You can erase, restore, and expand the canvas directly on the web, and use prompts together with selections to repaint specific areas. Below, following the real operation flow, I’ll clearly explain the key capabilities of the Midjourney Image Editor.

Where to access the Midjourney Image Editor and requirements

Open one of your works on the Midjourney web app, and you can usually see the “Edit” entry. Click it to enter the new Midjourney Image Editor interface. This editor also supports uploading images from your computer, then modifying and inpainting them in the same way.

Because the feature is still being rolled out in stages, some accounts may be restricted. Common thresholds include annual subscribers, users with a high cumulative generation count, and continuous monthly subscribers, etc. If you don’t see the Edit entry, first confirm you’re on the web app and that your account permissions include access to the Midjourney Image Editor.

Erase/Restore: use selections + prompts for localized repainting

The core of the Midjourney Image Editor is “region selection + text instructions.” You can use “Erase” to brush out the parts you don’t want, then clearly describe in the prompt what to fill in—for example, “change the cup in the hand into a camera, metallic texture, keep the lighting unchanged.”

If you erase too much, use “Restore” to brush the original area back and try again. It’s far more efficient than repeatedly using Vary. It’s recommended to add a line like “keep the overall composition and lighting” to the prompt, which can significantly reduce the chance of the change bleeding into other areas.

Canvas expansion and transforms: easier cropping, padding, and aspect-ratio changes

Previously, turning a landscape image into a portrait often meant generating from scratch; now, in the Midjourney Image Editor you can directly expand the canvas to add space above, below, left, or right. After expanding, combine it with localized repainting to have Midjourney fill in missing background—useful for leaving whitespace on covers, adding copy space on posters, or adjusting composition for e-commerce hero images.

You can also crop first and then inpaint: crop the subject tighter first, then expand the edges so the background becomes cleaner. For brand visuals, this “lock the composition first, then fill details” workflow is more controllable.

Retexturing and mixing references: style, character, and personalization together

The Midjourney Image Editor also supports a “retexturing” approach: preserve the scene’s shapes and structure as much as possible, then swap out the materials, surfaces, and lighting as a whole. You can change a “wooden table” into “white marble,” or a “daytime scene” into a “neon night scene”—the structure stays intact, but the look changes completely.

A practical tip is to mix style references with personalization: for example, include a style reference with --sref in the prompt while enabling the --p personalization model, so the Midjourney Image Editor follows your aesthetic while staying close to the reference style. If you also need character consistency, append --cref URL at the end of the prompt and use --cw to adjust the strength (the higher the value, the more it “resembles” the reference).