87 lines
4.9 KiB
Markdown
87 lines
4.9 KiB
Markdown
---
|
|
description: Edit/update an existing image using Gemini Flash via Nano Banana MCP with professional prompting
|
|
argument-hint: <image_path> <edit_description>
|
|
allowed-tools: [Read, Write, Glob, Grep, Bash, mcp__nano-banana-mcp__edit_image, mcp__nano-banana-mcp__continue_editing, mcp__nano-banana-mcp__get_last_image_info, mcp__nano-banana-mcp__get_configuration_status]
|
|
---
|
|
|
|
Edit/update an existing image based on the provided image path and edit description.
|
|
|
|
**Usage**: `/update <image_path> <edit_description>`
|
|
**Example**: `/update /heroes/camelot/photos/portrait.jpg add a mystical staff in his hand`
|
|
|
|
Follow these steps:
|
|
|
|
1. **Configuration Check**: First verify that Gemini API is properly configured using `mcp__nano-banana-mcp__get_configuration_status`
|
|
|
|
2. **Image Path Validation**:
|
|
- Parse the first argument as the image path
|
|
- Verify the image file exists and is accessible
|
|
- Extract the remaining arguments as the edit description
|
|
- If no image path provided, check if there's a recent image using `mcp__nano-banana-mcp__get_last_image_info`
|
|
|
|
3. **Context Discovery**: Based on the image location and edit description, gather relevant context:
|
|
- If image is in `/heroes/` directory, load character description
|
|
- If in `/locations/` or `/cities/`, load location context
|
|
- If in `/items/`, load item specifications
|
|
- If in `/creatures/`, load creature details
|
|
- Search for applicable settings in `/settings/` directory
|
|
- Load world information from `/worlds/` directory if applicable
|
|
|
|
4. **Entity Analysis**: Parse the edit description to identify:
|
|
- Changes requested (add, remove, modify elements)
|
|
- New elements to be introduced
|
|
- Style preferences or requirements
|
|
- Lighting or mood adjustments
|
|
- Technical specifications needed
|
|
|
|
5. **Edit Prompt Generation**: Create a professional edit prompt following `@docs/prompting-guide.md` editing templates:
|
|
- **Adding/Removing Elements**: "Using the provided image of [subject], please [add/remove/modify] [element] to/from the scene. Ensure the change is [integration description]."
|
|
- **Inpainting**: "Using the provided image, change only the [specific element] to [new element]. Keep everything else exactly the same, preserving the original style, lighting, and composition."
|
|
- **Style Transfer**: "Transform the provided image into [artistic style]. Preserve the original composition but render it with [stylistic elements]."
|
|
- **Detail Preservation**: "Using the provided image, [edit request]. Ensure that [critical details] remain completely unchanged. The modification should [integration requirements]."
|
|
|
|
6. **Settings Integration**: Apply appropriate settings for consistency:
|
|
- Maintain character clothing styles if character is present
|
|
- Preserve world-specific aesthetic elements
|
|
- Apply environmental settings for background consistency
|
|
- Include weapon/item styling if applicable
|
|
- Maintain magical effects consistency
|
|
|
|
7. **Reference Image Handling**: Determine editing approach:
|
|
- If editing a specific image file: use `mcp__nano-banana-mcp__edit_image` with the image path
|
|
- If continuing work on the last generated image: use `mcp__nano-banana-mcp__continue_editing`
|
|
- Include additional reference images if needed for style transfer or element addition
|
|
|
|
8. **Image Generation**: Execute the edit using the appropriate MCP tool:
|
|
- `mcp__nano-banana-mcp__edit_image` for specific file edits
|
|
- `mcp__nano-banana-mcp__continue_editing` for iterative improvements
|
|
|
|
9. **File Management**: Upon successful generation:
|
|
- Save updated image in the same directory structure
|
|
- Create versioned filename if preserving original
|
|
- Archive the successful edit prompt in corresponding `/prompts/` directory
|
|
- Update any cross-references if relationships changed
|
|
|
|
10. **Documentation**: Create a markdown file in `/prepared_prompts/` directory documenting:
|
|
- Original image path and edit description
|
|
- Final edit prompt used
|
|
- Applied settings and context
|
|
- Generated image path
|
|
- Edit type and template used
|
|
- Validation results
|
|
|
|
**Edit Types Supported**:
|
|
- **Element Addition**: Adding objects, characters, effects to existing image
|
|
- **Element Removal**: Removing unwanted elements while maintaining scene integrity
|
|
- **Modification**: Changing colors, expressions, poses, clothing, etc.
|
|
- **Style Transfer**: Converting image to different artistic style
|
|
- **Inpainting**: Replacing specific parts while preserving the rest
|
|
- **Composition Changes**: Adjusting framing, perspective, lighting
|
|
|
|
**Important**:
|
|
- The edit prompt must be in English regardless of input language
|
|
- Follow exact editing templates from `docs/prompting-guide.md`
|
|
- Preserve critical details mentioned in project files
|
|
- Maintain consistency with established character/location features
|
|
- Use appropriate MCP tool based on editing scenario
|
|
- Ensure proper file versioning to preserve original if needed |