Hero Image Generation

How It Works

During Brief Creation

1. Prompt Generation - ILLIXIS analyzes your keyword, title, content strategy, and brand context
2. Visual Context - Extracts your brand name, description, category, target audience, and image style preferences
3. Prompt Storage - Saves the AI-generated prompt to the brief for reuse

During Content Generation

1. Automatic Trigger - When article content is generated, hero image generation queues automatically
2. Async Processing - Image generation runs independently (won't block article generation)
3. Image Upload - Generated image uploads to secure storage with SEO-optimized filename

Result: Every published article has a professional hero image without manual work.

What Makes ILLIXIS Hero Images Different

Brand-Aware Generation

The system uses your tenant settings to ensure visual consistency:

• Brand Name & Description - Informs aesthetic and tone
• Target Audience - Ensures people shown reflect your audience authentically (age, body type, ethnicity, style)
• Image Style Description - Custom guidance on subjects, scenes, diversity preferences (see Configuration below)
• Primary Category - Influences visual style (fashion vs. tech vs. finance)

Composition Intelligence

Critical Design Rule: All images use center-weighted composition to work across multiple aspect ratios.

• 16:9 landscape - Original generation format
• 1:1 square - For Instagram, Facebook
• 9:16 vertical - For Stories, Reels, TikTok

By centering the main subject, images won't lose critical elements when cropped for social media.

Editorial-Quality Prompts

ILLIXIS generates 50-100 word prompts with specific details:

Includes:

• Lighting and mood (e.g., "golden hour," "soft natural window light")
• Composition details (e.g., "shallow depth of field," "overhead flat-lay")
• Color palette (e.g., "warm saturated colors," "earthy eco-conscious aesthetic")
• Atmosphere (e.g., "relaxed aspirational lifestyle," "professional editorial")

Excludes:

• Text, logos, watermarks, branded product shots
• Generic descriptions ("a nice picture of...")
• Stock photo aesthetics

Example Prompt:

"Golden hour photograph of a woman walking along a pristine Caribbean beach, wearing flowing resort wear. Shallow depth of field focuses on her silhouette against turquoise waves. Warm, saturated colors with soft highlights on water droplets. Relaxed, aspirational lifestyle mood."

Configuration

Basic Settings

Navigate: Settings > Visual Preferences

Allow People in Images (checkbox)

• Enabled: Images can include people
• Disabled: Images will not include people
• Default: Disabled (safer for broad use)

Advanced Settings

Image Style Description (text field) Define what should appear in your AI-generated images across all content.

What to include:

• Subject preferences (e.g., "lifestyle photography," "abstract concepts," "product closeups")
• People diversity (e.g., "show diverse body types, ages 25-45, multiple ethnicities")
• Scene types (e.g., "outdoor natural settings," "modern office environments")
• Aesthetic preferences (e.g., "bright and airy," "moody and dramatic")

Example:

"Show diverse people (ages 25-45, multiple body types and ethnicities) in active lifestyle settings. Prefer outdoor natural scenes with bright natural lighting. Avoid corporate stock photo aesthetics. Include authentic moments and real expressions."

This description gets passed to every hero image prompt generation, ensuring brand consistency.

Viewing Generated Prompts

During Brief Creation:

1. Create a Keyword or Custom Brief
2. Wait for brief generation to complete (3-5 minutes)
3. View brief details - the hero image prompt is saved in the brief

During Content Generation:

1. Generate article from brief
2. Image generates automatically in background (10-15 seconds)
3. View content - featured image displays with generated prompt stored in metadata

Regenerating Hero Images

If the generated image doesn't match your vision, regenerate with a custom prompt.

Navigate: Content Hub > View Content > Image tab

Options:

Option 1: Use Existing Prompt (Retry)

Click "Regenerate Image" without changing the prompt. This sends the same prompt again, producing a different result.

Use when: The prompt is good but the specific image generated isn't quite right.

Option 2: Edit Prompt

Modify the existing prompt or write a new one from scratch, then click "Regenerate Image."

Use when: You want more specific visual details or a different concept entirely.

Option 3: Regenerate from Brief

Delete the current image and regenerate the article. The system will use the original brief prompt again.

Use when: You want to return to the original AI-generated prompt.

Prompt Priority Logic

When generating hero images, the system checks these sources in order:

1. Custom prompt provided - If you manually enter a prompt during regeneration
2. Existing image prompt - AI-generated prompt from previous generation
3. Brief image prompt - AI-generated prompt from brief creation
4. Fallback prompt - Simple prompt built from title/keyword (rare)

This priority ensures consistency while allowing manual overrides.

Image Processing

Format: PNG generated, converted to JPEG (80-90% smaller file size) Rate Limit: 18 requests per minute Retry Logic: 3 attempts with exponential backoff Timeout: 90 seconds

Unsplash Image Availability Monitoring

If you use Unsplash images (hotlinked from Unsplash's CDN), ILLIXIS monitors them for availability.

Why images become unavailable:

• Photographer removed the image from Unsplash
• DMCA takedown request
• Unsplash policy violation
• Account deletion

Daily monitoring: ILLIXIS automatically scans all content with Unsplash images at 3:00 AM UTC, checking each URL with an HTTP HEAD request. If any images are unavailable, a digest email is sent at 4:00 AM UTC listing affected content.

Digest email contents:

• List of content pieces with broken Unsplash images
• Direct links to each affected article
• Recommended action: Replace with AI-generated image or upload your own

How to replace broken Unsplash images:

1. Open the content piece from the digest link
2. Navigate to Image tab
3. Either:

• Click "Generate AI Image" to create a replacement
• Upload your own replacement image

4. Save changes

Prevention strategies:

• Use AI-generated images instead of Unsplash for permanent content
• Enable "Auto-download Unsplash" in Settings > Visual Preferences (stores local copy)
• For critical content, always use uploaded or AI-generated images

Image Generation Queue

• Processes continuously (no scheduled window)
• Images queue immediately when content generation completes
• Typical processing time: 10-15 seconds per image

Retry Behavior

• Failed generations automatically retry after 5 minutes
• Maximum 3 retry attempts before marking as failed
• Exponential backoff between retries (5 min, 10 min, 20 min)
• Failed images can be manually regenerated from Content Hub

Sibling Handling

If the content has a bilingual sibling (e.g., EN/ES versions), the hero image is automatically copied to the sibling to avoid generating duplicate images for the same article.

Troubleshooting

Image Not Generating

Check these first:

1. API Key Missing

• Verify image generation API credentials are configured in Settings > Integrations

2. Rate Limit Hit

• Symptom: Image generation queues but doesn't complete
• Fix: Wait 60 seconds for rate limit window to reset

3. Image Generation Disabled

• Symptom: Image generation is skipped
• Fix: Check Settings > Visual Preferences to confirm image generation is enabled

4. Task Timeout

• Too many concurrent image requests
• Fix: Automatic retry will succeed after backoff

Image Doesn't Match Brand

Problem: Generated images don't reflect your brand aesthetic.

Fix: Configure Image Style Description in Settings > Visual Preferences.

Example configurations:

Tech SaaS:

"Modern office environments with diverse professionals (ages 25-45). Clean minimalist aesthetic. Bright natural lighting. Show people collaborating, using laptops. Avoid stock photo poses."

Fashion Brand:

"Lifestyle photography featuring diverse models (sizes 0-14, ages 18-40, multiple ethnicities). Outdoor natural settings. Golden hour lighting. Authentic movement and expression. Avoid studio backgrounds."

Food Blog:

"Overhead flat-lay food photography. Natural ingredients on wooden surfaces. Soft diffused lighting. Show hands preparing food. Earthy, organic aesthetic. Avoid commercial kitchen settings."

Prompt Too Generic

Problem: ILLIXIS generates vague prompts like "a nice beach picture."

Likely cause: Missing content strategy or brief context.

Fix: Ensure briefs have:

• Target keyword set
• Title generated
• Content strategy present (from SERP analysis or AI insights)

If brief has all context but prompt is still generic, regenerate the brief (rare AI response issue).

Image Cropped Wrong on Social

Problem: Important visual elements cut off when used in social posts.

Cause: Composition wasn't center-weighted.

Fix: Regenerate with explicit composition instruction:

"Center the main subject in the middle third of the frame. Avoid placing critical elements near left/right edges. Composition must work when cropped to square (1:1) or vertical (9:16)."

Common Workflows

Standard Article Flow

1. Create Keyword Brief - hero prompt auto-generated
2. Generate article - image auto-generated from prompt
3. Publish - article has hero image

No manual steps.

Custom Image for Specific Article

1. Generate article with default image
2. Navigate to Content Hub > View Content > Image tab
3. Enter custom prompt (or edit existing)
4. Click "Regenerate Image"
5. New image replaces old one

Update Brand Visual Guidelines

1. Settings > Visual Preferences
2. Update "Image Style Description" field
3. Save
4. All NEW briefs will use updated guidelines
5. Existing briefs keep their original prompts (regenerate briefs to apply new guidelines)

Best Practices

DO:

• Configure Image Style Description before creating briefs
• Use specific visual details in custom prompts (lighting, composition, mood)
• Center important subjects for social media cropping
• Let ILLIXIS generate the first prompt (it has more context than you realize)

DON'T:

• Request text or logos in images (the AI blocks these)
• Use vague prompts ("a nice picture of X")
• Place key visual elements at frame edges
• Regenerate images unnecessarily (each regeneration counts toward your quota)

Related Features

• Inline Image Placeholders - Strategic image placeholders in article body (separate from hero images)
• Banner Ad Generation - Display ads with AI backgrounds
• Quote Graphics - Social media quote cards with AI backgrounds
• Email Header Images - Email campaign headers with AI-generated visuals

All visual generation features use the same brand context and image style settings for consistency.