Midjourney User Guide: Generate images with the Web version and manage your gallery to download the originals

This article clearly explains how to use Midjourney’s Web version to generate images, edit them, organize your work in the gallery, and then download the original files. The workflow starts with logging in and linking accounts, then moves through prompt input, upscaling and variations, and finally exporting files. As long as your Midjourney account has an active subscription/permissions, you can follow along and complete the whole process.

1. Sign up, log in, and link your account—open up the access path first

Open the Midjourney official website and log in; common methods include email or third-party authorized sign-in. The first time you use it, the page will usually prompt you to link Discord to sync your historical creations and permission status. If you previously only used Midjourney in Discord, after linking, the Web side will automatically pull in your generation history.

After linking, it’s recommended to quickly check your profile page: whether your username, avatar, and linking status are all normal. If you run into a blank page after logging in or an endless loading spinner, try an incognito/private window first; then clearing the site cache usually resolves it.

2. Start generating on the Web: prompt input and basic settings

After entering the creation page, you can generate with Midjourney by typing your prompt directly into the input box. It’s recommended to write descriptions in the order “subject + style + lighting + lens/composition + detail constraints.” Short phrases can also work, but the clearer the information, the more stable the results. If you have reference images, you can upload them first, then use the image link together with the text prompt.

The Web version generally provides common options like aspect ratio/size and style strength. Running one round with the default values is the most hassle-free. If you want more control, lock in one size and one style strength, and afterward only change the subject description—this makes Midjourney’s results easier to compare and iterate.

3. Upscale, variations, and re-rendering: turn “usable” into “what you want”

After generation, you’ll usually see entry points for upscaling and variations: upscaling is for getting a clearer single image, while variations are for continuing to explore with a similar composition. It’s recommended to upscale the one that’s closest to your goal first, then do variations; otherwise you can easily waste time bouncing between four small thumbnails. When you need local changes, use the re-render/inpainting (regional edit) feature: circle the area you want to change and add additional instructions, and Midjourney will recompute that part according to your command.

If the style drifts as you keep editing, go back to the earliest “baseline image” and generate variations from it again for more stable results. You can also add descriptions of “elements you don’t want” in the prompt to reduce repeated re-rendering.

4. Gallery management and downloading originals: organize, retrieve, and export in one go

Midjourney’s gallery page supports filtering by time, task, and image status. It’s recommended to favorite key works, which will save you a lot of effort when searching later. To download the original image, open the single image detail page first, then choose the download option—this helps you avoid accidentally saving a browser-compressed preview. For sharing publicly, a share link is clearer than screenshots and also makes it easier to trace back to the corresponding generation task.

If you find that some works are missing in the gallery, first confirm you’re logged into the same Midjourney account, and check whether you’ve switched to a different identity/workspace. Syncing between the Web and Discord records can sometimes be delayed; refreshing the page or logging out and back in usually restores missing items.