Midjourney Vary Region Command Guide – Changing and Adding Elements

Midjourney continues to push the boundaries of image editing with their latest breakthrough: the Vary Region editor. In August 2023, they unveiled this remarkable feature, allowing users to selectively enhance and modify specific sections of their images without altering the entire picture.

You can use the Vary Region to enhance your image with new components, eliminate unwanted elements, use reference images, and mix things up in a totally fresh way, all without changing the parts of your image that you want to remain the same.

It’s the newest image editing tool in your design arsenal and we’ve put together a Midjourney Vary Region Command Guide to show you how to use it!

The Midjourney Vary Region Command Guide – Changing and Adding Elements

While Midjourney’s Vary Region tool may seem like a game-changer in the world of image editing, mastering its subtleties is the key to unlocking its full potential. Without a grasp of the nuances, you might not achieve the results you’re aiming for.

We’ve made this guide to demystify the Vary Region editor and show you how to harness its power to transform your images like a pro.

How to Change An Element In An Image With The Vary Region Command

If you have an image with an element you don’t like, we’re going to show you how to change it.

Note: If you’re looking to add an element to an image that was not there before, keep reading! Because afterwards, we are going to show you how to do that as well.

Step 1: Create an Image.

Make yourself an image in Midjourney using the /imagine command.

Prompt: a Doberman Pincer portrait, wearing a top hat, photorealistic, studio lighting

I love everything about image #1, except for that darned bow tie.

If you find yourself with an image like this one that would be almost perfect, except for one element, you could re-run the prompt another 150 times and hope to get an image as good as the one you liked, without the random element in it.

Or you can save yourself the time and disappointment and use the new Midjourney Vary Region editing tool. 

If you want to go for option 2 – follow the steps below.

Step 2: Upscale the image.

Once you’ve upscaled your image you will see the Vary Region button underneath. Click it!

Step 3: Select the area you want to change.

You can either use the square selection tool, or the lasso tool to select the area you want to change. For the square selection tool, click and hold the mouse button while dragging to select the area.

If you make a mistake, you can use the circular arrow ‘undo’ icon on the top left, and try your selection again.

Step 4: Press Submit.

What’s going to happen is that Midjourney will change only what is in the selected area. Also note that it will give you different variations of what’s in the selected area.

So, in this case it just gave me a bunch of Dobermans with different bow ties, as opposed to different things around its neck.

If the dog had been wearing a scarf in the original image, it would have given me the dog with different scarves around its neck. Also note that it only gave different versions of a bowtie, but not much of a variation in color.

In some instances, depending on the image and what you want to change, the straight-up Vary Region tool on its own might give you what you’re looking for.

However, if you are looking to change your selected area with something completely different (or even just a different color), then you are going to have to adjust your prompt.

If you noticed, when selecting the Vary Region button and making the selection, the prompt didn’t come up and there was no way to change it.  

In order to use the Vary Region tool to replace your selected area with something specific, you will need to use the Remix mode.

Step 5: Activate Remix Mode.

To activate the Remix mode, type in /settings and hit enter. A box will pop-up. Click the Remix mode button and make sure it’s green.

Step 6: Go back to your upscaled image and click Vary Region again.

At this point you will need to find that image you upscaled earlier. With people on Discord constantly generating images and the feed scrolling a mile a minute, it can be a bit of a nightmare trying to find an image you created earlier, especially if it wasn’t in the last 5 minutes. A good tip: start DMing directly with the Midjourney Bot!

Step 7: Select the area you want to change and adjust the prompt.

Now that you’ve activated the Remix mode and once again found your original upscaled image that you wanted to change and clicked on the Vary Region button, you’ll notice something different. Not only will you be able to select the area you want to change, but now you will also see, and be able to edit, the prompt.

Because you are only editing that small area, you can delete the original prompt and replace it with only what you want to see in that area.

For my edited prompt, I simply put: “dog collar –no bow tie.”

Now it may sometimes seem that people are creating these master works of art and design at the click of a button when you read all the Midjourney “how to” or “prompts for..” articles. But what you don’t ever see is the hundred and fifty re-runs of the image using different prompt tweaks, parameter settings and editing modes to get the image they used.

So, don’t worry if you don’t get the image you wanted on the first try. You will almost never get the image you wanted the first time around. If you did, there wouldn’t be a need for all of those different settings and editing modes.

For my first go round, this is what I got:

I wanted a dog collar. And I guess technically they are dog collars, but that’s not what I had in mind… at all. At this point you have one of two options. If you’ve used the Remix mode to alter the prompt and you’re not getting what you wanted, you can:

1. Re-run the original upscaled image in with Vary Region in Remix mode and finetune your prompt to make it more specific, like: “black leather dog collar, silver buckle –no bow ties”;


2. Use a reference image

I did try to re-run the image a few more times, fine tuning the prompt and nothing I got was what I wanted. In fact, sometimes Midjourney completely ignored the prompt and gave me some pretty random results, along with more bow ties despite my having used the –no bow ties command.

Step 8: Try A Reference Image

I did a Google image search for dog collars and found one that was what I had envisioned the first time around and another one that I thought I might try…

In order to use a reference image, you first need to upload it to Discord.

For an image on the web, you can right click the image and select copy. Go to Discord and right click in the text window and select paste. Press enter to upload your image.

For an image on your hard drive, you simply drag the image into the text field and press enter. Or you can also press the + sign to the left of the text field and select “upload file” from the pop up.

Once your image is uploaded into the Discord feed, right click it and select “copy link.” Paste that link into Word, Notepad or a similar text program – you’re going to need that link in a minute.

Step #9: Go back to your original upscaled image & click Vary Region

Don’t forget – you’ll find an image you’ve created a lot faster by looking through the ‘mentions’ tab in your inbox, than trying to scroll and find it.

Once you’ve found that original upscaled image on the feed, and have clicked Vary Region again, copy the link from the image you uploaded earlier and paste it into the prompt window.

Note: you cannot re-run an image with just a reference image link – you must also add a prompt. Just leave a space and write a short description of the object, then press submit.

I ran the Vary Region editor twice with the different collars separately. This is what I got:

To be honest I wasn’t super impressed with any of the renditions above. You might have a little more luck depending on the image you choose to alter.

Ironically, despite it not having initially been what I’d had envisioned when I wanted to change the bow tie, the best “non-bow tie” image (in my opinion) came out of the first run, without tweaking the prompt or using a reference image.

At the moment, the Vary Region command can be a bit of a hit-and-miss, depending on your image and what you want to do to it, but you have to remember this is a new feature and Midjourney is always being updated.

How to Add An Element To An Existing Image With The Vary Region Command

Now we’re going to show you how to, not change, but add an element to your image.

Step 1: Create an image using the /image command.

This will work with any image you create.

Step 2: Upscale your image and select the Vary Region button.

Once you choose the image that you like from the grid, upscale it and click the Vary Region button. Make sure you are in the remix mode. If you’ve jumped ahead and skipped over the above sections, there are detailed instructions on activating the remix mode here. [anchor link to step #5 above]

Step 3: Select the area where you want to add something to your image, and add a prompt.

I chose U4 to upscale the 4th image in my grid, selected the area I wanted to add something and added the prompt: “local fishing boat.” And pressed the submit/arrow button to the right of the text field.


And now you have four views of your image with your added element. If you are not satisfied with the results, you can always use a reference image of the item you want to add to your image.

If you skipped over it above, you can find detailed instructions on how to upload and add a reference image to the Vary Region editor here. [add anchor link to Step #8:try a reference image]

Take Away

The Vary Region Command is an image editing game changer, making the already amazing Midjourney AI text-to-image generator even more flexible.

That’s said, the Vary Region command is a relatively new feature, so much like all of the other features that have rolled out from Midjourney, it may take a little while for this new function to get up to speed with the rest of the program.

For more advanced Midjourney tecniques, see some of our other articles like: