Analyzes an attached character image to perform deep feature extraction (facial physiognomy, physical attributes, costume details) and generates a split-screen image containing two full-body views (Front and Back) on a simple color background, maintaining the exact visual style of the source.
-
Research Character (Visual & Web Search):
- Visual Analysis: detailed feature extraction (physiognomy, attributes, costume).
- Web Search: If the character appears to be from a franchise or has a name, use to find setting/lore details to inform dynamic poses.
-
Construct the Image Generation Prompt:
- Layout & Views:
- Default (Split Screen): "Split-screen, Left: Front view, Right: Back view."
- 2x2 Layout: "Four-panel grid. Top-Left: Front view. Top-Right: Back view. Bottom-Left & Bottom-Right: Dynamic action poses [based on character setting]."
- 3x3 Layout: "Nine-panel grid. Center: Front view. Top-Center: Back view. Remaining panels: Varied dynamic poses and close-ups [based on character setting]."
- Mandatory: Front and Back views are always required.
- Background: "...on a simple color background."
- Details: "Exact mirror of facial features, physical attributes, and costume details from the source."
- Visual Constraints: "Style Fidelity: Retain the exact visual style, rendering technique, and lighting quality of the source image."
-
Generate the Image:
- Use the image generation tool with the constructed prompt and the user's reference image.
User: "Create a reference sheet for this character." (User attaches
)
Action:
User: "Make a 3x3 reference sheet for 'Cloud Strife'." (User attaches
)
Action: