How to: Generate consistent AI characters across images with Midjourney

Rock Trembath
11 Jun 202403:22

TLDRDiscover how to create and maintain consistent AI characters across multiple image generations with Midjourney's latest update. This tutorial guides you through generating a full-body character image, using it as a reference for various scenes in a storyboard. Learn to upscale images, utilize the 'cref' command for character consistency, and iterate through descriptions to achieve a cohesive look across different scenarios, enhancing the visual communication for your team.

Takeaways

  • 🚀 Midjourney's update allows for easier generation of consistent AI characters across multiple images.
  • 🎨 It's recommended to start with a full body view of the character for consistency in various scenes.
  • 💡 Including details like time period, clothing, and pose in the imagine prompt helps in creating the desired character.
  • 🔍 A full body, wide shot with a white background is suggested for the initial character image to avoid anomalies in image generation.
  • 📸 After generating the initial character, upscaling and using it as a reference for subsequent images is crucial.
  • 🔗 Utilize the image URL from the initial character to maintain consistency in the MidJourney channel.
  • 📝 Describe the scene and use the 'cref' command with the character reference to generate images for specific scenarios.
  • 🤔 There can be challenges with character placement and anatomy, which might require adjustments in the imagine prompt.
  • 🔄 Iteratively refining the imagine prompt and using the cref command can improve the accuracy of the character in different scenes.
  • 👔 Consistency in facial features and clothing can be achieved more effectively with the update compared to previous versions.
  • 👨‍👩‍👧‍👦 The generated images can be used for storyboarding and visual communication within a team, even if not perfect.
  • 🌐 Encouragement to try Midjourney, share creations, and stay creative is a part of the video's closing message.

Q & A

  • What update has Midjourney released to facilitate character consistency across images?

    -Midjourney has released an update that makes it significantly easier to create a character and then use it across multiple instances of new image generations.

  • Why is a full body frontal view of a character recommended when creating a reference for Midjourney?

    -A full body frontal view is recommended because it includes all necessary body parts such as legs and arms, which are important for various scenarios and helps to prevent the generation of weird or incorrect body parts.

  • What are the key elements to include in an imagine prompt for creating a character in Midjourney?

    -Key elements include the character's description, time period, hair color, clothing, body posture, and background.

  • How does the video demonstrate the process of creating a consistent character for a storyboard?

    -The video demonstrates by creating a full body view of a 'middle aged suburban dad' and then using that image as a reference for subsequent image generations with the 'cref' command to maintain consistency.

  • What command is used in Midjourney to maintain character consistency when generating new images?

    -The 'cref' command is used to maintain character consistency by referencing a previously generated image.

  • What challenges might one face when trying to generate consistent images of a character in Midjourney?

    -Challenges may include incorrect body part placement, changes in clothing, and variations in facial features that do not match the reference image.

  • How can one improve the accuracy of character consistency in Midjourney?

    -One can improve accuracy by adjusting the imagine prompt to be more specific, such as adding details about the character's pose or the scene, and by trying multiple generations to find the best match.

  • What is the purpose of using a character reference in Midjourney's image generation process?

    -The purpose of using a character reference is to ensure that the character's appearance remains consistent across different image generations, which is useful for creating storyboards or maintaining a character's identity in a series of images.

  • Can the Midjourney update guarantee perfect consistency between renders?

    -The update significantly improves consistency, but it may not guarantee perfect consistency between renders, as some variations can still occur.

  • How can one share their renderings with the creator of the video?

    -Viewers can share their renderings with the creator on Twitter, as suggested in the video.

  • What is the final advice given by the video creator to the viewers?

    -The final advice is to stay creative, try out the Midjourney update, and not forget to share their creations.

Outlines

00:00

🎨 Character Creation for Image Generations

The script introduces a new update from Midjourney that simplifies the process of creating and reusing characters in various image generations. The narrator suggests starting with a full body view of the character to avoid inconsistencies in future scenes. The example given is creating a middle-aged suburban dad for a commercial storyboard, with specific details like the 1990s time period, brown hair, and a striped tie. The character is generated using an 'imagine prompt' and then upscaled for clarity. The process emphasizes the importance of including the entire body in the reference image to ensure accurate generation in different scenarios.

Mindmap

Keywords

Midjourney

Midjourney is the name of the AI tool discussed in the video, which is used for generating images based on textual prompts. In the video, the creator uses Midjourney to generate a consistent character across multiple images, which is the main theme of the tutorial. The tool's capabilities are showcased through the process of creating and referencing a 'middle-aged suburban dad' character.

Character Reference (cref)

Character Reference, or 'cref', is a command used within the Midjourney tool to ensure that the generated images maintain consistency with a previously created character. In the script, the creator uses 'cref' to reference an initial image of the character, aiming to generate new images of the same character in different scenes.

Image Generation

Image generation refers to the process of creating visual content from textual descriptions using AI. In the context of the video, the creator generates images of a character by inputting descriptive prompts into the Midjourney tool, which then produces images that match the description.

Storyboard

A storyboard is a sequence of images or illustrations displayed in a comic strip format, showing the progression of events in a script. In the video, the creator is using Midjourney to generate images for a storyboard, which will visually communicate the scenes of a commercial.

Upscale

To upscale an image means to increase its resolution while maintaining or improving its quality. In the video, the creator upscales the initial character image to enhance its detail before using it as a reference for further image generation.

Full Body View

A full body view in image generation refers to creating an image that includes the entire body of the subject, from head to toe. The video emphasizes the importance of a full body view to avoid anomalies when generating images for different scenarios.

Middle-aged Suburban Dad

This term describes the character that the creator wants to generate images of in the video. It is a specific archetype that includes attributes like age, location, and possibly lifestyle, which are used in the imagine prompt to guide the AI in generating the character's image.

1990s

The '1990s' is a time period mentioned in the video that serves as a stylistic and cultural reference for the character and scenes being generated. It helps to set the aesthetic and thematic context for the images produced by Midjourney.

Imagine Prompt

An imagine prompt is a textual description provided to the AI tool to guide the image generation process. In the video, the creator uses various imagine prompts to describe the character and scenes, which the AI then uses to create the corresponding images.

AI Characters

AI characters are virtual personas created by AI tools like Midjourney. They are not real people but can be designed to appear lifelike. In the video, the creator discusses how to generate and maintain consistency of an AI character across different images.

Consistency

Consistency in this context refers to the ability to generate images of the same character with similar features and attributes across multiple instances. The video demonstrates how Midjourney's update allows for greater consistency in character representation in different scenes.

Highlights

Midjourney's update simplifies the creation and reuse of AI characters across multiple image generations.

Creating a full body view of the character is recommended for consistency in various scenarios.

Including limbs and full body in the initial character image prevents generation anomalies.

Using an 'imagine prompt' to describe the character for the first image generation.

Example: Creating a 'middle-aged suburban dad' character for a storyboard.

Describing the character with attributes like '1990s', 'brown hair', and 'striped tie'.

The importance of a 'full body, wide shot' for the initial character image.

Upscaling the initial character image for better detail.

Using the image URL as a character reference for subsequent image generations.

Describing new scenes with commands like '1990s, midshot, man sitting in car'.

Utilizing 'cref' to maintain character consistency across image generations.

Adjusting descriptions to improve character consistency, such as 'staring forward through windshield'.

Accepting minor inconsistencies for storyboard purposes.

Iteratively refining the 'cref' command to improve character matching.

Observing improvements in facial features and clothing consistency in newer model versions.

Encouraging users to experiment with Midjourney and share their renderings.

Emphasizing the value of creativity in the use of AI character generation.