Batch Image Generator

Batch Image Generator

This skill generates multiple images with consistent visual style for use in video production. It manages the batch generation process, ensures style coherence, and organizes output assets.

When to Use This Skill

• Creating multiple B-roll images for video
• Generating consistent visual assets for video series
• Producing background images for digital human videos
• Creating graphics and illustrations for explainer videos
• Generating product shots or scene backgrounds
• Building visual asset libraries for projects

What This Skill Does

1. Batch Processing: Generates multiple images efficiently
2. Style Consistency: Ensures visual coherence across all images
3. Quality Control: Validates outputs against requirements
4. Organization: Saves and categorizes generated assets
5. Metadata Tracking: Records prompts and settings for each image
6. Iteration Management: Handles regeneration and refinements

How to Use

From Prompt List

Generate images from these prompts: [filename or paste list]

From Storyboard

Generate all images needed for storyboard: [storyboard filename]

Batch Specification

Generate [N] images with consistent style:
- Style: [description]
- Theme: [topic]
- Purpose: [use case]

Instructions

When user requests batch image generation:

1. Understand Requirements

   I'll generate a batch of images for your video.
   
   **Prompt Source**:
   [ ] Prompt file (already have prompts)
   [ ] Storyboard (generate from storyboard)
   [ ] Description (I'll create prompts)
   
   **Generation Platform**:
   [ ] Midjourney
   [ ] DALL-E 3
   [ ] Stable Diffusion
   [ ] Multiple platforms
   [ ] Your choice (I'll recommend)
   
   **Batch Size**: [Number of images]
   
   **Style Requirements**:
   - Visual style: [Photorealistic/Illustrated/3D/etc.]
   - Color palette: [Specific colors or mood]
   - Consistency level: [Strict/Moderate/Varied]
   
   **Timeline**: [Urgent/Standard/Flexible]
   

2. Load or Create Prompts

   ## Prompt Preparation
   
   **Total Images**: [N]
   **Style Template**: [Base style to apply to all]
   
   ### Prompt List
   
   **Image 001**: [Name/Description]
   

   [Full prompt]

   **Negative**: [Negative prompt]
   **Purpose**: [Where used in video]
   
   **Image 002**: [Name/Description]
   

   [Full prompt]

   **Negative**: [Negative prompt]
   **Purpose**: [Where used in video]
   
   [Continue for all images...]
   

3. Style Consistency Check

   ## Consistency Validation
   
   **Common Style Elements** (present in all prompts):
   - ✅ Color palette: [Specified in all]
   - ✅ Lighting style: [Consistent]
   - ✅ Quality keywords: [Same across all]
   - ✅ Technical specs: [Matching]
   - ✅ Artistic style: [Consistent]
   
   **Variable Elements** (changes per image):
   - Subject matter
   - Specific composition
   - Scene details
   
   **Potential Issues**:
   - [ ] No style conflicts detected
   - [ ] Minor variations noted: [List]
   - [ ] Significant inconsistencies: [List and recommend fixes]
   

4. Platform Selection & Optimization

   ## Platform Recommendation
   
   **Recommended Platform**: [Platform name]
   
   **Reasoning**:
   - [Why this platform is best for these images]
   - [Cost consideration]
   - [Quality expected]
   - [Generation speed]
   
   **Alternative Options**:
   - **Option 2**: [Platform] - [Pros/Cons]
   - **Option 3**: [Platform] - [Pros/Cons]
   
   **Platform-Specific Optimizations Applied**:
   - [Optimization 1]
   - [Optimization 2]
   - [Optimization 3]
   

5. Generation Plan

   ## Batch Generation Plan
   
   **Strategy**:
   [ ] Sequential (one at a time, review each)
   [ ] Parallel (batch submit, review all together)
   [ ] Hybrid (key images first, then batch)
   
   **Grouping**:
   
   **Group 1: High Priority** (Generate first)
   - Image 001: [Description] - [Reason for priority]
   - Image 003: [Description] - [Reason for priority]
   
   **Group 2: Standard**
   - Image 002: [Description]
   - Image 004: [Description]
   - Image 005: [Description]
   
   **Group 3: Optional/Backup**
   - Image 006: [Description]
   - Image 007: [Description]
   
   **Estimated Timeline**:
   - Group 1: [X minutes]
   - Group 2: [X minutes]
   - Group 3: [X minutes]
   - Total: [X minutes to X hours]
   

6. Generation Execution

   ## Generation Log
   
   ### Image 001: [Name]
   **Status**: ⏳ Generating... / ✅ Complete / ❌ Failed / 🔄 Regenerating
   **Prompt Used**: `[Prompt]`
   **Platform**: [Platform name]
   **Attempts**: [N]
   **Generation Time**: [X seconds/minutes]
   **File**: `./output/assets/images/generated/image_001.png`
   
   **Quality Check**:
   - [ ] Matches prompt description
   - [ ] Style consistent with batch
   - [ ] Technical quality sufficient
   - [ ] Resolution correct
   - [ ] Usable for video
   
   **Issues**: [None / List any problems]
   **Action**: [Approved / Regenerate / Adjust prompt]
   
   ---
   
   ### Image 002: [Name]
   [Same structure as Image 001]
   
   ---
   
   [Continue for all images...]
   
   ---
   
   ## Batch Summary
   
   **Generated**: [X] / [Total]
   **Success Rate**: [X]%
   **Failed**: [N] - [Reasons]
   **Regenerations Needed**: [N]
   

7. Quality Control Process

   ## Quality Control Checklist
   
   ### Per-Image QC
   
   For each image:
   - [ ] Subject matter correct
   - [ ] Composition as specified
   - [ ] Style matches batch
   - [ ] Colors align with palette
   - [ ] Technical quality high
   - [ ] No artifacts or errors
   - [ ] Resolution sufficient (min [X]p)
   - [ ] Aspect ratio correct
   - [ ] Usable for intended purpose
   
   ### Batch-Level QC
   
   - [ ] Visual consistency across all images
   - [ ] Color palette coherent
   - [ ] Style uniformity maintained
   - [ ] Quality level consistent
   - [ ] No outliers or mismatches
   
   ### Issues Found
   
   **Image [N]**: [Issue description]
   - **Severity**: [Low/Medium/High]
   - **Action**: [Approve anyway / Minor edit / Full regenerate]
   - **Note**: [Any relevant notes]
   
   **Image [N]**: [Issue description]
   [Same structure]
   

8. File Organization

   ## File Management
   
   **Directory Structure**:
   

   /output/assets/images/generated/[project-name]-[date]/ ├── image_001_[description].png ├── image_002_[description].png ├── image_003_[description].png ├── ... ├── batch_metadata.json ├── prompts_used.txt └── generation_log.md

   
   **Naming Convention**:
   - Format: `image_[###]_[short-description].png`
   - Example: `image_001_office-workspace.png`
   
   **Metadata File** (batch_metadata.json):
   ```json
   {
     "batch_id": "[unique-id]",
     "project": "[project-name]",
     "date": "[date]",
     "total_images": [N],
     "platform": "[platform]",
     "style": "[style-description]",
     "images": [
       {
         "id": "001",
         "filename": "image_001_[description].png",
         "prompt": "[full-prompt]",
         "negative_prompt": "[negative-prompt]",
         "generation_time": "[timestamp]",
         "attempts": [N],
         "purpose": "[video-scene]",
         "status": "approved"
       },
       // ... more images
     ]
   }
   

9. Regeneration Management

   ## Regeneration Queue
   
   **Images Requiring Regeneration**:
   
   ### Image [N]: [Name]
   **Reason**: [Why regenerating]
   **Original Prompt**: `[prompt]`
   **Adjusted Prompt**: `[new-prompt]`
   **Changes Made**: [Specific adjustments]
   **Attempt**: [N]
   **Status**: [Pending / In Progress / Complete]
   
   ### Image [N]: [Name]
   [Same structure]
   
   ---
   
   **Regeneration Strategy**:
   - **Variation level**: [Same with tweaks / Significant changes]
   - **Expected attempts**: [N]
   - **Time estimate**: [X minutes]
   

10. Cost & Resource Tracking

    ## Resource Usage
    
    **Generation Costs** (if applicable):
    - Platform: [Platform name]
    - Images generated: [N]
    - Images kept: [N]
    - Total attempts: [N]
    - Cost per image: [Amount]
    - Total cost: [Amount]
    - Credits used: [N]
    - Credits remaining: [N]
    
    **Time Investment**:
    - Prompt creation: [X minutes]
    - Generation waiting: [X minutes]
    - QC and review: [X minutes]
    - Regenerations: [X minutes]
    - Total time: [X hours]
    

11. Batch Delivery Report

    ## Batch Generation Complete! ✅
    
    **Project**: [Project name]
    **Date**: [Date]
    **Total Images**: [N]
    
    ### Deliverables
    
    **Generated Images** ([N] files):
    - Location: `./output/assets/images/generated/[folder]/`
    - Format: PNG
    - Resolution: [X x Y]
    - Total size: [X] MB
    
    **Documentation**:
    - `batch_metadata.json` - Full generation data
    - `prompts_used.txt` - All prompts (copy-ready)
    - `generation_log.md` - Detailed generation log
    
    ### Quality Metrics
    
    - Success rate: [X]% first attempt
    - Images approved: [N]
    - Images regenerated: [N]
    - Average attempts: [X]
    - Style consistency: [High/Medium/Low]
    
    ### Image Breakdown by Purpose
    
    | Purpose | Count | Images |
    |---------|-------|--------|
    | Backgrounds | [N] | [List IDs] |
    | B-roll | [N] | [List IDs] |
    | Products | [N] | [List IDs] |
    | Graphics | [N] | [List IDs] |
    | Other | [N] | [List IDs] |
    
    ### Style Consistency Report
    
    **Color Palette Adherence**: [X]%
    **Lighting Consistency**: [X]%
    **Quality Level**: [Consistent / Minor variations / Some outliers]
    **Overall Visual Coherence**: [Score or description]
    
    ### Issues & Notes
    
    **Challenges Encountered**:
    - [Issue 1]: [How resolved]
    - [Issue 2]: [How resolved]
    
    **Recommendations**:
    - [Recommendation for future batches]
    - [Platform insights]
    - [Prompt refinements]
    
    ### Next Steps
    
    1. Review all images: `Show me all generated images`
    2. Regenerate specific image: `Regenerate image [N] with [changes]`
    3. Generate variations: `Create 3 variations of image [N]`
    4. Use in video: `Create video using these images`
    5. Generate more: `Generate [N] more images in same style`
    
    **Ready for video production!** 🎬
    

12. Create Usage Guide

    ## Image Usage Guide
    
    ### For Video Editing
    
    **Importing Images**:
    ```bash
    # Import all images
    ls ./output/assets/images/generated/[folder]/*.png
    

    Recommended Usage:

    Image 001 (office-workspace.png):

    • Use for: B-roll during intro section
    • Duration: 3-5 seconds
    • Timing: 0:15-0:20 in video
    • Transition: Fade in from previous shot

    Image 002 (product-closeup.png):

    • Use for: Feature demonstration
    • Duration: 4-6 seconds
    • Timing: 0:45-0:50 in video
    • Transition: Cut from talking head

    [Continue for all images...]

    Color Grading Notes

    To Match Video:

    • Temperature: [Adjustment]
    • Tint: [Adjustment]
    • Saturation: [Adjustment]
    • Contrast: [Adjustment]

    LUT Recommendation: [If applicable]

13. Save All Files

    • Images: ./output/assets/images/generated/[project]/
    • Log: ./output/assets/images/generated/[project]/generation_log.md
    • Metadata: ./output/assets/images/generated/[project]/batch_metadata.json
    • Prompts: ./output/assets/images/generated/[project]/prompts_used.txt

Output Format

Batch generation should produce:

• High-quality images: Meeting technical requirements
• Consistent style: Visually coherent set
• Organized files: Properly named and categorized
• Complete documentation: Prompts, metadata, logs
• Ready to use: Immediately usable in video production

Tips

• Test first: Generate 2-3 samples before full batch
• Consistency tokens: Use same seed or style parameters when possible
• Version prompts: Small tweaks can cause big style changes
• Backup strategy: Generate extras for critical shots
• Batch efficiently: Group similar subjects together
• Document everything: Save prompts that work well
• Quality over quantity: Better to regenerate than settle

Platform-Specific Tips

Midjourney

• Use --seed for consistency across batch
• Same --style value for all images
• Reference first image: --sref [URL]

DALL-E 3

• Keep prompts similar structure for consistency
• Use style references in all prompts
• Consider generation style (vivid vs natural)

Stable Diffusion

• Lock seed for consistency
• Use same model/checkpoint
• Same LoRA weights if applicable

Related Skills

• Use prompt-generator to create prompts first
• Reference project-context for brand style
• Input from storyboard-generator for shot list
• Output to video-segment-creator or video-compositor