Image Generation Tutorial

15 min read All Users Updated Jan 5, 2025

Create stunning storyboard images with AI. Learn prompt optimization and batch generation techniques for professional results.

Image Generation Basics

AI Storyboard uses advanced GLM image generation models to create high-quality storyboard frames from your descriptions. The AI interprets your scene parameters, camera settings, and artistic directions to generate visually consistent images.

Single Frame Generation

Generate One Frame at a Time

  1. Select a frame from the scene list
  2. Review and adjust parameters if needed
  3. Click "Generate Image" button
  4. Wait for generation (typically 10-30 seconds)
  5. Review the generated image
  6. Regenerate if needed by clicking "Regenerate"

When to Use Single Generation

  • Fine-tuning specific scenes
  • Testing different prompts
  • When each scene has unique requirements
  • Limited network bandwidth

Batch Generation

Generate Multiple Frames

To generate multiple frames at once:

  1. Select multiple frames (or select all)
  2. Click "Batch Generate Images"
  3. Monitor progress in the status bar
  4. Images generate sequentially
  5. Review each image as it completes

Batch Generation Tips

  • Start with small batches (3-5 frames)
  • Ensure consistent style settings across batch
  • Monitor for any frames that need regeneration
  • Save your project before large batches

Prompt Optimization

Effective Prompt Structure

A good image generation prompt includes:

1. Subject - Main focus of the image

A young female detective with ponytail, wearing gray trench coat

2. Action - What the subject is doing

examining evidence, holding flashlight

3. Environment - Setting and context

abandoned warehouse, dim light from high windows, dusty atmosphere

4. Camera - Shot type and angle

medium shot, slight low angle, dramatic shadows

5. Style - Artistic direction

cinematic, film noir style, moody lighting, high contrast

Optimization Techniques

Be Specific

Good:

Detective in warehouse, examining clues, flashlight beam cutting through dust,
dramatic rim lighting, cinematic, film noir

Too vague:

Detective looking at stuff

Use Professional Terminology

Incorporate film and photography terms:

  • Lighting: rim light, fill light, backlit, high-key, low-key
  • Composition: rule of thirds, centered, dutch angle
  • Focus: shallow depth of field, bokeh, rack focus
  • Style: cinematic, photorealistic, dramatic, moody

Order Matters

Structure your prompt with most important elements first:

  1. Main subject
  2. Action/interaction
  3. Environment
  4. Camera details
  5. Lighting
  6. Style modifiers

Troubleshooting

Images Not Matching Expectations

Solution:

  • Review your prompt for clarity
  • Add more specific details
  • Include camera and lighting information
  • Try regenerating with same prompt (AI varies each time)
  • Adjust style keywords

Inconsistent Style Across Frames

Solution:

  • Use consistent style keywords in all frames
  • Maintain same lighting descriptions
  • Keep color palette references similar
  • Use "Consistency Settings" for characters and locations

Generation Taking Too Long

Solution:

  • Check internet connection
  • Reduce batch size
  • Generate during off-peak hours
  • Close unnecessary browser tabs

Advanced Tips

1. Iterative Refinement

  • Generate initial image
  • Identify what needs improvement
  • Adjust prompt based on results
  • Regenerate with refined prompt

2. Reference Frames

  • Generate one "hero frame" with perfect style
  • Use it as reference for other frames
  • Copy style keywords to maintain consistency

3. Batch Size Strategy

  • Start with 3-5 key frames
  • Review and adjust based on results
  • Generate remaining frames in batches
  • Fine-tune individual frames as needed