How to Use Image Prompts and the /blend Command in Midjourney

Words only get you so far. When you have a reference image, you can paste its URL into a prompt so the model borrows its colors, composition, or subject. The /blend command takes this further by merging two to five uploaded images into a single new picture without any text at all.

What you need

• Two to five reference images on your device
• An active Midjourney subscription
• A channel or DM where you can upload files

Step 1: Upload an image to get its URL

Drag an image into the Discord message box and send it on its own. Click the uploaded image, choose Open in Browser, and copy the address. That link is now a usable image reference.

Step 2: Put the URL at the front of /imagine

In a new /imagine prompt, paste the image URL first, add a space, then add your text description and parameters. The image influences color and composition while your words steer the subject.

/imagine prompt: https://cdn.discordapp.com/.../ref.png a sports car in the same color palette, studio lighting --ar 16:9

Step 3: Run /blend for a pure image merge

Type /blend and select it. Discord presents image slots labeled image1 through image5. Drop a photo into each slot you want to use. No text prompt is needed, which makes blend perfect for quick mashups.

Step 4: Choose dimensions and submit

Open the optional dimensions field and pick Square, Portrait, or Landscape to set the aspect ratio of the blend. Press Enter and the bot returns a grid that fuses your inputs.

Step 5: Upscale your favorite blend

Treat the blend output like any grid. Click a U button to upscale, then Open in Browser to save the full-resolution file.

Example: blending a moody portrait with a cracked-paint texture produces a stylized character whose skin carries the texture, something that would be hard to describe in words alone.