How To Create Multiple AI Characters In One Scene

Playground AI
5 May 202410:22

TLDRIn this tutorial, the speaker demonstrates how to create a scene with multiple AI characters without merging their traits, a common issue when generating images of multiple characters together. The process involves using a platform called 'canvas' to compose the scene with a background and three characters, removing the background from each character, and adjusting their size and position for a natural look. To address lighting inconsistencies, an 'image to image' technique is applied with a sunset lighting effect. The characters' details are then manually enhanced through a process of cropping and upscaling, ensuring better facial features and hand details. The final step involves composing the refined images into a single scene with consistent lighting and details, producing a well-controlled and detailed final image. The speaker also mentions an upcoming update to the platform, Playground V3, which will integrate canvas and board features.


  • 🎨 **Scene Composition**: To create a scene with multiple AI characters, you start by composing the scene with a background and individual characters.
  • 🧩 **Avoiding Character Merging**: Prompting for two characters together might result in a merge; using separate elements helps maintain distinct identities.
  • πŸ“ **Canvas Familiarity**: Ensure you are comfortable with the canvas tool before starting to avoid complications.
  • πŸ’Ύ **Save Originals**: Keep the original character images as a backup in case you need to start over.
  • πŸ—œ **Removing Backgrounds**: Remove the background from each character image for a clean composition.
  • πŸ”„ **Scene Duplication**: Duplicate the scene to introduce characters and adjust their size and position.
  • πŸŒ„ **Lighting Consistency**: Address lighting differences between characters and background by using image-to-image adjustments.
  • πŸ” **Detail Tweaking**: After lighting adjustments, manually upscale and tweak details like hands and facial features for a more refined image.
  • πŸ“ˆ **Manual Upscaling**: Use a manual upscaling process to control the composition and improve details without affecting the entire image.
  • πŸ–Œ **In-Painting Tool**: Use the in-painting tool to seamlessly blend areas that don't match the background.
  • ♻️ **Iterative Process**: The process is iterative, requiring recomposing, cropping, and upscaling individual characters to achieve the final scene.
  • πŸ”§ **UI Update Note**: The author mentions an upcoming update (Playground V3) that will combine canvas and board, which may change the UI but retain the same tools and concepts.

Q & A

  • What is the main purpose of creating multiple AI characters in one scene?

    -The main purpose is to avoid the merging of character traits that can occur when prompting for multiple characters in a single image. By creating each character separately and then composing them into a scene, one can achieve a more accurate and distinct representation of each character.

  • Why is it recommended to keep the original character assets when starting the process?

    -It is advised to keep the originals in case things go wrong or if you need to redo a step. This ensures you have a backup to revert to and try again without starting from scratch.

  • How does removing the background from each character help in the scene composition process?

    -Removing the background from each character allows for greater flexibility in composing the scene. It enables the characters to be placed naturally within the scene without the distraction of their original backgrounds.

  • What is the significance of adjusting the lighting in the composed image?

    -Adjusting the lighting is crucial because it ensures that the lighting is consistent across all subjects in the scene, which can otherwise look unnatural if the lighting on the subjects differs from the background.

  • How does the 'image to image' feature help in achieving the desired lighting effect?

    -The 'image to image' feature allows the user to apply a specific type of lighting, like sunset lighting, to the entire scene. It helps in creating a cohesive look where the lighting on the characters matches the lighting of the background.

  • What is the reason for manually upscaling individual characters after composing the scene?

    -Manual upscaling is done to improve the details of each character without affecting the overall composition. It allows for better control over the final image quality, ensuring that details like facial features and hands are clear and well-defined.

  • Why is it important to crop tightly around each character during the upscaling process?

    -Tight cropping ensures that only the character is upscaled, maintaining the original composition and background details. It prevents unnecessary changes to the background and helps in focusing the upscaling on the character alone.

  • How does changing the image strength during the 'image to image' process affect the final image?

    -Changing the image strength allows for control over how much of the original composition is retained versus how much of the new lighting effect is applied. A higher strength retains more of the original image, while a lower strength allows more of the new lighting to be visible.

  • What is the benefit of using the in-painting tool to fix blurry areas in the composed image?

    -The in-painting tool helps to seamlessly integrate the upscaled characters into the background, removing any visible seams or blurry areas that might detract from the overall image quality.

  • Why is it necessary to regenerate areas of the image that do not match the background?

    -Regenerating these areas ensures that the final image is cohesive and that the characters appear naturally within the scene. It helps to avoid any inconsistencies that could make the image look unrealistic or poorly composed.

  • What is the advantage of using the method described in the transcript for controlling the scene and characters?

    -The method allows for precise control over the composition, enabling the creator to place specific characters in the scene exactly as desired. It also provides a way to achieve higher quality images by pulling in more details during the upscaling process.



🎨 Composing a Scene with Multiple Characters

The video begins with an introduction to creating a scene with multiple characters using a tool called 'canvas'. The speaker explains the limitations of AI models when trying to generate images of two characters together, such as Batman and Superman, which often results in a merged, inconsistent look. To overcome this, the use of canvas is suggested to manually compose the scene. The process involves selecting and removing backgrounds from individual character images and then arranging them in a scene with a background. The importance of keeping original images is emphasized in case of errors. The video also touches on adjusting the size and position of characters and the need for consistent lighting across the scene.


πŸŒ„ Adjusting Lighting and Image Upscaling

The second paragraph delves into addressing lighting inconsistencies in the composed scene. The speaker demonstrates how to use an 'image to image' feature with a sunset lighting prompt to harmonize the lighting across all characters and the background. The process includes copying and pasting the scene into the image to image tool, adjusting the image strength, and fine-tuning the prompt to achieve the desired lighting effect. After establishing consistent lighting, the focus shifts to manually upscaling individual character images to improve details like facial features and hands. This is done by cropping the characters tightly, ensuring shadows are retained, and using the image to image tool again with a focus on enhancing details. The speaker also discusses the process of refining the upscaled images to fit seamlessly into the scene.


πŸ“ˆ Final Scene Composition and Future Updates

In the final paragraph, the speaker wraps up the scene composition process by emphasizing the iterative nature of the work, which may involve regenerating images to achieve a seamless blend between characters and the background. The use of the eraser tool and in-painting feature is highlighted for touching up areas that need refinement. The video concludes with a mention of upcoming updates to the playground V3, which will feature a new UI combining canvas and board functionalities. The speaker also points to other videos that cover utilizing drawings to further control and condition images, before signing off.



πŸ’‘AI Characters

AI Characters refer to artificial intelligence entities that are created and controlled by software. In the context of the video, the creator is discussing how to generate and manipulate multiple AI characters within a single scene using a digital canvas, which is a tool for designing and composing images.

πŸ’‘Scene Composition

Scene Composition is the arrangement of visual elements within a given space to create a coherent and aesthetically pleasing image. In the video, the creator is focusing on how to compose a scene with a background and multiple characters, ensuring that they are well-integrated and visually appealing.

πŸ’‘Background Removal

Background Removal is the process of digitally separating a subject from its background, typically for the purpose of placing the subject against a different background or for a transparent backdrop. The script describes how the creator removes the background from each character to prepare them for the scene composition.

πŸ’‘Image to Image

Image to Image is a technique used in digital image editing where one image is used as a reference to modify another. In the video, the creator uses this technique to apply consistent lighting to the characters, making them match the background's lighting conditions.

πŸ’‘Lighting Consistency

Lighting Consistency refers to the uniformity of lighting across different elements in an image or scene. The video emphasizes the importance of achieving consistent lighting to make the composite scene look natural and believable.

πŸ’‘Manual Upscaling

Manual Upscaling is the process of increasing the resolution of an image while manually enhancing the details to maintain or improve the image quality. The creator discusses this process as a way to refine the details of the characters after they have been scaled up.


In-Painting is a digital image editing technique used to fill in or smooth out selected areas of an image, often to remove unwanted elements or to blend edges. In the video, the creator uses in-painting to clean up the edges of the characters and to correct areas where the background is blurry.


Canvas, in the context of digital art and design, refers to a virtual space where images and other visual elements can be arranged, edited, and composed. The video is a tutorial on using a canvas tool to create a scene with multiple AI characters.

πŸ’‘Control Net

Control Net is a term that might refer to a feature or tool within an image editing software that allows for precise control over the editing process, such as the manipulation of specific elements within a scene. The creator mentions utilizing Control Net to achieve the desired results in the scene.

πŸ’‘Playground V3

Playground V3 is a reference to an upcoming version of a digital platform or software that the creator mentions is in development. It is implied to be an improved or updated version of the current tool, which will combine features like canvas and board for a more integrated user experience.


Drawings, in the context of the video, are mentioned as another method to condition and control images, suggesting that the creator also discusses or will discuss the use of hand-drawn or digital sketches as part of the image creation process.


The video demonstrates how to create multiple AI characters in one scene using a composition technique to avoid character merging.

A limitation of current AI models is that prompting for multiple characters may result in a merge of their traits.

Utilizing canvas allows for more control over the final composition of characters in a scene.

The process involves removing the background from individual character images to ensure clear separation.

Duplicating the scene and manually adjusting the size of characters helps in composing the desired scene layout.

The importance of keeping original character assets in case of errors or the need to retry is emphasized.

Lighting consistency across subjects and background is crucial and can be achieved through image-to-image adjustments.

Image strength between 45 to 50 is suggested as a starting point for image-to-image lighting adjustments.

Manual upscaling is preferred for controlling the original composition while improving details.

Cropping individual characters tightly, including shadows, is essential before upscaling.

Using the 'paint' tool for in-painting can help in seamlessly blending regenerated areas with the original image.

Changing the prompt during image regeneration to include the subject improves the output.

The final scene composition involves lining up new assets and making adjustments for consistency.

The process, though time-consuming, allows for detailed control over the scene and character placement.

Playground V3 is in development, which will combine canvas and board for an improved user interface.

Drawings can be used as another method to condition and control images in the scene.

The video provides a comprehensive guide for users new to canvas, assuming some familiarity with the tool.

The presenter shares their personal approach to scene composition for those who want specific control over their AI-generated scenes.