The Editing an Image with AI feature in Layer allows you to refine and modify existing images using powerful language models like ChatGPT 4o. This feature is perfect for making precise adjustments to characters, assets, and scenes without starting from scratch, while maintaining the consistency of the original design.
Feature Overview
There are many use cases for this feature, but here are some of the most impactful ways it can be used, along with example prompts and expected results:
Color Adjustment
Easily change the color of characters, objects, or environments without altering their overall design.
Example Prompt: “Change the color of the gem on the ring to purple, but keep the rest of the ring’s design, material, and shape exactly the same.” | |
Expected Result: The gem color updates to a rich purple hue, while the ring’s shape, material, and design details remain completely unchanged. |
Upgrade Chains
Guide the AI to progressively enhance assets, perfect for multi-stage items like treasure chests, tools, or character equipment.
Example Prompt: “Create a jewelry item set that upgrades from level 1 to level 10. Each level should increase in size, complexity, material quality, and number of items—starting with a simple ring and ending with an elaborate crown, earrings, necklace, and bracelet.” |
Expected Result: A complete jewelry set evolves across 10 levels. Early levels show simple, small pieces (e.g. a plain gold ring), and each level introduces more intricate detailing, additional gemstones, new materials like platinum or enchanted metals, and eventually a full matching set of ornate, high-status accessories.
Perspective Shift
Generate different views of characters or objects, like rotating a front-facing character to a side profile or rear view, without changing their proportions or style.
Example Prompt: “Change the character from a forward-facing view to a three-quarter side profile, but keep the proportions and style consistent.” | |
Expected Result: The character is rotated to show a more dynamic, three-quarter angle, retaining the original design and proportions. |
Element Removal
Clear items from a scene for before/after comparisons, while keeping the remaining elements unchanged.
Example Prompt: “Remove all the furniture from this room, but keep the lighting, walls, and overall style exactly the same.” | |
Expected Result: The scene is cleared of all furniture, leaving a clean, open space while maintaining the original style. |
Adding Props or Accessories
Attach reference images to guide the AI for complex edits, like adding weapons, clothing, or other accessories.
Example Prompt: “Make the character hold the sword and shield from the attached reference image. Keep the character’s pose, proportions, and overall design the same, only adding the sword and shield to the character’s hands.” | |
Expected Result: The character is updated to include the sword and shield, positioned naturally in their hands without altering the character’s overall design. |
Style Matching for New Elements
Ensure added props, accessories, or characters match the original art style for a cohesive look.
Example Prompt: “Add a penguin sidekick to this character. Have the penguin standing on their shoulder, and match the penguin’s style to the same lighting, shading, and proportions as the rest of the image. Change nothing else about the image.” | |
Expected Result: A stylized penguin is added to the character’s shoulder, seamlessly matching the character’s armor style, lighting, and the overall aesthetic of the snowy mountain scene. The rest of the image remains untouched. |
Background Integration
Modify or extend backgrounds to better fit new characters or assets without disrupting the original setting.
Example Prompt: “Add a stone pathway leading to the house, but match the lighting, color tones, and perspective of the existing background.” | |
Expected Result: A stone pathway appears, naturally blending into the scene with consistent perspective and lighting. |
Pose Adjustments
Make small adjustments to a character’s pose without altering their overall style or proportions.
Example Prompt: “Update the character’s pose so that both hands are in his pockets. Don’t change the character’s design, the angle of the image, or the overall image art style.” | |
Expected Result: The character’s arms are repositioned so that both hands are casually placed in his pockets. The lighting, angle, expression, outfit, and art style remain exactly the same, ensuring a seamless pose update without disrupting the scene. |
Character Expression Updates
Change a character’s expression without modifying their overall design, perfect for adding emotional variety.
Example Prompt: “Make this character have a sad, crying expression. Change nothing else about the character or the overall image—keep the pose, lighting, and style exactly the same.” | |
Expected Result: The character’s facial expression is updated to look visibly sad and teary-eyed, while the rest of the image remains completely unchanged in style, pose, color, and composition. |
How to Use the Editing Feature
Access the Editing Tool:
Right-click on any asset in your Layer workspace.
Select Refine → Prompt Edit from the context menu.
Select Your LLM:
Choose an LLM like ChatGPT 4o, depending on the style and detail level you need for your edits.
Enter the Prompt or Attach a Reference:
Text Prompt: Type a specific instruction, like the examples above, to guide the AI’s edits.
Reference Image: Attach a reference image for complex edits, like adding props or changing poses, for more precise results.
Prompting Tips
Be Direct: Clearly instruct the AI on the specific change you want, like “Add a shield to the character’s left arm,” rather than just describing the scene.
Preserve Consistency: Remind the AI to keep the original style, shape, and overall design intact for the parts of the image you don’t want to change.
Use References for Precision: For complex changes, like adding accessories or altering poses, reference images can significantly improve accuracy.
Think in Steps: For upgrade chains or multi-phase designs, break the edits into clear steps for consistent results.