How to Maintain Character Consistency in Midjourney!
When using Midjourney to generate character images, a significant challenge is maintaining character consistency! This is because Midjourney is an AI diffusion model that creates images from noise, which introduces a high level of randomness. Even when the same command is input, the output results can vary each time.
Today, we will introduce five methods to maintain character consistency in Midjourney.
Step 1: Generate a Character with Keywords
In Midjourney, input keywords to create a character you like, select your preferred image, and choose U3.
Descriptive terms: Blind box style, Cute little girl with rabbit ears hat, Full body, Look at the camera, Pop mart, octane rendering, ultra details, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K
Step 2: Copy Image Link
Select your preferred image, click U3 to enlarge it, right-click to get the URL (link) and copy it.
Step 3: Remember the Universal Formula: Image URL + Keywords + iw 2
In the command box, input “/imagine” followed by the URL at the beginning of the descriptive terms, and add –iw2 at the end.
Descriptive terms: image link Blind box style, Cute little girl with rabbit ears hat, Full body, Look at the camera, Pop mart, octane rendering, ultra details, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K –iw 2
Step 4: Compare Results
You can compare the original image with the generated image this time. We can see that this method produces a high level of character consistency, and the style and overall atmosphere are basically consistent.
This method can generate a series of images using one image.
Step 1: Generate a Character with Keywords
In Midjourney, input keywords to create a character you like, select your preferred image, and choose U3.
Descriptive terms: Blind box style, Cute little girl with rabbit ears hat, Full body, Look at the camera, Pop mart, octane rendering, ultra details, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K
Step 2: Find the Image Seed Value
In the generated grid interface, click the emoji icon at the top right. In the search box, enter: envelope, find the envelope icon, and click it.
Step 3: Copy the Seed Value
At this point, you will receive a message from the Midjourney Bot. Open the message in your private chat, and you will obtain the seed value. Copy the seed value of the image.
Step 4: Remember the Universal Formula: Replace Keywords + Seed Value
Modify your keywords by changing “ears hat” to “ears hat”, and add –seed 1668584462 at the end.
Descriptive terms: Blind box style, Cute little girl with cat ears hat, Full body, Look at the camera, Pop mart, octane rendering, ultra details, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K –seed 1668584462
Step 5: Compare Results
The results are still very good. Come and give it a try! Of course, this is also a beginner’s method, and we will share some advanced techniques next.
This method can help create continuous actions.
1. Generate Different Actions: N panels with different poses
We input “4 panels with different poses + descriptive terms” to obtain character images with different actions.
Descriptive terms: 4 panels with different poses, Blind box style, Cute little girl with cat ears hat, Full body, Look at the camera, Pop mart, octane rendering, ultra details, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K
2. Generate Continuous Actions: N panels with continuous doing
We input “4 panels with continuous dancing + descriptive terms” to obtain a series of character images of dancing actions.
Descriptive terms: 4 panels with continuous dancing, Blind box style, Cute little girl with cat ears hat, Full body, Look at the camera, Pop mart, octane rendering, ultra details, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K
3. Generate Multi-Angle Actions: Character Sheet
When we input “character sheet with multiple poses and expressions + descriptive terms” we can obtain character images with multiple angles, various poses and expressions.
Descriptive terms: character sheet with multiple poses and expressions, a super cute girl, dreamy cute hair accessories, pop mart, blind box, octane rendering, edgelighting, chibi, The best quality, HD, C4D, chiaroscuro, 8K
Emojis represent expressions, and an Express sheet is a combination of various expressions. By using Midjourney’s “emoji + expression sheet,” you can easily create a series of character expressions.
Descriptive terms: 3D emoji, expression sheet, Super cute girl, IP by popmart, Disney style, Pixar style, Scenes in spring, pastel color, surrounded by flowers, fine luster, gendered composition, 3D render, Soft focus, blender
Descriptive terms: 3D emoji, expression sheet, a cute girl, furry clothes, full body, pop mart, chibi, C4D, octane rendering, edgelighting, depth, ultra details, The best quality, HD –ar 1:1
This method can produce images that resemble a series of movie storyboards in one go, with multiple consecutive images.
Introducing the universal formula: “Magic Prefix + Keyword Group + Image Ratio 9:16”
1. Beginning: Descriptive Terms
“children’s illustration”
“random film stills of DreamWorks animation”
“Cartoon animation character”
2. Middle: Descriptive Terms
Add “Continuous Shooting” or “Sequence shot” or “front side back three view” at any position in the middle, or all three, to ensure a 100% chance of generating storyboards.
3. Ending: –ar 9:16
Descriptive terms: animation style, Cartoon animation character, random film stills of DreamWorks animation, front side back three view, fantasy anime environment, the full body 3D artwork of a cute Chinese girl, Chinese style, in the style of romantic soft focus and ethereal light, sandara tang, oshare kei, study 3D game art, fisheye lens, studio light the focal length of the background is 35mm f1.4 –ar 9:16 –s 400
Method 1: Base Image Method
“Image URL” + Keywords + “ –iw 2”
Method 2: Seed Value Method
“Replace Keywords” + “–seed value”
Method 3: Panels Action Method
Generate Different Actions: N panels with different poses to generate different actions.
Generate Continuous Actions: N panels with continuous doing.
Generate Multi-Angle Actions: character sheet.
Method 4: Emoji Expression Method
“emoji” + “expression sheet (combination of various expressions).”
Method 5: Shoot Lens Method
“children’s illustration / random film stills of DreamWorks animation / Cartoon animation character” + “Continuous Shooting / Sequence shot / front side back three view” + “–ar 9:16”
If you want to learn AI, you can long press the QR code below to consult a professional teacher.
In addition to the above content, we have also prepared a free AI monetization resource package of over 10G+, which is continuously updated. After scanning the QR code below and following the steps, you can receive it:
