Midjourney Beginners #2: Power Tips for Image Editing with the Web Editor

How to edit Midjourney images like a pro using simple techniques

“Midjourney Beginners” is a series of articles written for readers who are just getting started with Midjourney.

Anyone can make beautiful images with a few simple steps.

No complicated things here.

---

Editing images in Midjourney

According to David, Midjourney is an image generator, not an editor.

But for many of us, it's a little bit of both.

If you only need to make minor changes or remove junk from an image before posting it to your social media account, you can do it the Midjourney web editor. Don't waste your money on the Adobe Photoshop subscription.

There are two image editors: (1) the Web Editor, which is available to everyone, and (2) the Full Editor, also known as the External Editor, which is available to users who have already generated 10,000 images or have an annual subscription.

The main difference between the two is that Full Editor allows users to edit any image, including those not created by Midjourney. Whereas Web Editor can only edit images created by Midjourney.

To keep things simple, I'll only cover the Web Editor, which is accessible to all beginners. Mind you, the "power tips" I'm about to share with you are handy and practical, even for intermediate or advanced users. I use them all the time in my image creation process.

---

Fixing common problems

1. Fix cropping issues

   1. The bot may occasionally cut off your subject's limps or a portion of their head. This is an easy fix. I will show you two simple solutions.

2. Add or remove things

   1. The bot may add "junk" to the image, such as an unexplained object or strangely colored areas. Removing them is simple with the Web Editor.

   2. Adding new elements to the image is more challenging, and the bot will fight against you and resist change. Well, just like in real life, not everyone likes change. Consider this: the bot proudly shows you its creation, and you say you want to add this and that. Why can't you add them to the prompt in the first place? So the bot is unhappy, and nothing will be changed.

   3. However, there is a trick that will make things easier. Although not a perfect solution, it will undoubtedly save you a lot of fast hours!

3. Multiple edits at once

   1. Many beginners are unaware that you can add and remove items at the same time in one run, saving you time and hours.

4. Add and edit text

   1. Midjourney text is a pain, especially when compared to other image generators, such as Ideogram.

   2. Midjourney V6.1 is still not very good at generating a large number of text-rich images. It may get it right on occasion, but it frequently makes a lot of mistakes. I'll give you some pointers on this topic.

---

(1) Fix cropping issues

a portrait photograph of a chubby cat wearing kimono, Japanese --sref 831214820 --ar 16:9 --v 6.1

Awwwwhhhhh…. cutie pie!

The image is almost purr-fect, except that the right front paw of the kitty has been cropped. Now let’s fix this issue.

Method A: Use the size bars (handles) to resize and click the aspect ratio

Resize and maintain the image’s aspect ratio

1. On the Midjourney website, click the Editor button. Then, using the size bars or handles, resize the image so that the paw fits inside the canvas.

2. To maintain the original image's overall aspect ratio, click on an aspect ratio that is similar to it. Then click Submit.

---

Method B: One-Click Zoom Out

Tip: Click the Zoom buttons 1.5x or 2x to zoom out and automatically correct the cropping issue in a click.

Zoom out is a quick way to correct the image cropping issues.

You can change the Zoom Out size using a custom value other than the standard 1.5x and 2x.

---

(2) Add or remove things

It is easier to remove elements from an image. However, adding new things will only work if they make sense to the bot.

It depends on the context.

For example, this prompt generates an image of a cat wearing a kimono. The image is about Japanese culture, geisha, and kimono attire.

a portrait photograph of a chubby cat wearing kimono, Japanese --ar 16:9 --v 6.1

As a result, even if you add a BIG element, such as an umbrella or samurai sword, the bot will most likely add it because of the context.

If you add a gold necklace, the bot will resist the addition because it is out of context with Japanese culture (p.s. necklaces are generally not recommended with kimono), so nothing will change.

Tip: Don't waste your fast hours by repeatedly running the edit to add an out-of-context element. The bot will win. You lost your fast hours.

Image 1: Original image. Image 2: Erase the area and enter the prompt "holding an umbrella". Image 3: Nice umbrella. Image 4: Holding a samurai sword, with an additional edit.

The bot will resist adding the out-of-context element. In this case, it's the gold necklace. Geisha typically wear minimal or no jewelry on the neck.

---

(3) Multiple edits at once

You can simultaneously delete and add new elements.

There is no need to do it in two steps i.e., remove a thing first, then use the prompt to add a new element.

A focused guy seated at a desk, their face illuminated by the soft glow of a computer screen displaying an image editing interface The overall color scheme is mainly pink, yellow, orange, purple, green, red, and black. In the style of Hayao Miyazaki anime style. --ar 16:9 --v 6.1

Areas 1 (an abnormal item on the right arm to be removed) and 2 (an image of the cat to be added to the screen) can be edited together by erasing both areas and using the prompt "illustration of a cute cat." The bot is smart and will only add the cat image on the screen.

Tip: Save your fast hours and time. Combine multiple edits in a single run.

The result.

---

Add and edit text

Midjourney V6.1 is not very good at texting. So, we'll have to be creative to work around the limitations.

Method A: Add and remove text in one run (low success rate)

Similar to multiple edits, you can try your luck with this method. Sometimes, it works perfectly. Other times, give up if nothing good comes out after 4-5 re-runs.

Remove abnormal text at the corner and add text “LOVE”. Prompt: a portrait photograph of a chubby cat wearing kimono, Japanese --sref 831214820 --ar 16:9 --v 6.1

Sometimes you get good result in a run. But this is rare.

---

Method B: Use Niji model

The Niji model handles text better.

It is easier to add text than Midjourney.

Although the model does not render the speech bubble, the generated text is correct and easier to create. Prompt: a portrait photograph of a chubby cat wearing kimono, Japanese --sref 831214820 --ar 16:9 --niji 6

You can switch to the Niji model by selecting it from the Version menu.

In addition to Midjourney V6.1, there are Niji models and useful older versions.

---

Method C: Begin with an image containing text. Edit the text to create new words.

Add the text to your prompt. Then edit the text to correct the spelling or make up a new word. The original word "Kitty" has been changed to "Love".

a portrait photograph of a chubby cat wearing kimono, Japanese, "Kitty" --ar 16:9 --sref 831214820 --v 6.1

Tip: Avoid erasing the original text completely. Leave a trace behind as a "lead" for the bot to generate new text.

The trick is not to completely erase the original text before creating a new word. If you did that, the bot would most likely remove everything and not generate any words.

The result.

---

Related articles

The Complete Guide to Midjourney Image Editor

Midjourney Beginners #1: How to steal the look of a masterpiece

---

Key Learnings

• Midjourney offers two image editors: the Web Editor (available to all) and the Full Editor (for advanced users). The Web Editor can only edit Midjourney-generated images.

• Common editing techniques include

  • Fixing cropping issues by resizing or using zoom-out options

  • Adding or removing elements that fit the image context

  • Performing multiple edits simultaneously to save time

• Text editing in Midjourney V6.1 is challenging. Workarounds include:

  • Attempting to add/remove text in one run (low success rate)

  • Using the Niji model for better text handling

  • Starting with an image containing text and editing it

• When editing, it's important to consider the image context and the AI's limitations. The bot may resist some changes if they don't align with the original concept.

---

---