Generate Interior Moodboards with Claude/Gemini Agents, Hugging Face Image Generation, and PDF Export

🎨 Template Overview

This comprehensive n8n workflow automates the complete process of generating professional interior design moodboards from concept to client delivery. Users submit a design brief through a form, and the system automatically generates 12 AI-powered images, compiles them into a beautifully formatted two-page PDF moodboard, and emails the final deliverable.

Key Features: Form-based design brief submission AI-powered image prompt generation (12 detailed prompts per project) Automated image generation via Hugging Face API Nextcloud cloud storage with public URL sharing Professional two-page HTML/PDF moodboard creation Automated email delivery with PDF attachment

Technologies Used: Claude Sonnet 4 (OpenRouter), Google Gemini 2.5 Pro, Hugging Face FLUX.1-schnell, Nextcloud, Gotenberg PDF Service, Gmail

⚙️ Self-Hosted Requirements

This template requires the following self-hosted or third-party services:

Nextcloud Instance** - For cloud file storage and public URL generation Gotenberg PDF Service** - For HTML to PDF conversion (can be self-hosted via Docker) OpenRouter API Access** - For Claude Sonnet 4 AI agent Google Gemini API Access** - For secondary AI processing Hugging Face API Access** - For FLUX.1-schnell image generation Gmail Account** - For email delivery (or any SMTP service)

📋 Setup Instructions

Step 1: Configure API Credentials

OpenRouter (Claude Sonnet 4) Sign up at openrouter.ai Generate an API key Add credentials to the "OpenRouter Chat Model" node

Google Gemini API Visit Google AI Studio Create an API key Add credentials to the "Google Gemini Chat Model1" node

Hugging Face API Register at huggingface.co Generate an access token from Settings → Access Tokens Add credentials to the "Image Generator" node using HTTP Header Auth Header name: Authorization, Value: Bearer YOUR_TOKEN

Nextcloud Set up a Nextcloud instance or use a hosted provider Generate an app password from Settings → Security Configure credentials in all Nextcloud nodes: "Create a folder" "Upload Image" "Share a file"

Gotenberg PDF Service Self-host using Docker: docker run --rm -p 3000:3000 gotenberg/gotenberg:8 Or use a hosted instance Update the URL in the "PDF creator" node Configure HTTP Basic Auth credentials if required

Gmail Enable 2-Factor Authentication on your Google account Generate an App Password from Google Account settings Add OAuth2 credentials to the "Send PDF" node

Step 2: Customize Workflow Settings

Email Extractor Node Review the stripPlus variable (default: true) This removes "+tag" portions from email addresses for folder naming

Nextcloud Folder Structure Default path: /moodboard/{username}/ Modify in "Create a folder" node if needed

Image Generator Settings Model: FLUX.1-schnell (fast generation, good quality) Adjust model in "Image Generator" node if needed Alternative models: FLUX.1-dev, Stable Diffusion XL

PDF Generation Settings Default timeout: 360 seconds Page size: A4 (210mm × 297mm) Adjust in "PDF creator" node headers if needed

Step 3: Test the Workflow

Activate the Form Open the "Moodboard Form" node Copy the webhook URL Access the form in your browser

Submit a Test Request Fill in the form fields: Title: Short, descriptive name (e.g., "Modern Minimalist Bedroom") Description: Detailed design brief with colors, materials, mood, lighting Email: Your test email address Submit and monitor workflow execution

Verify Each Stage Check Nextcloud folder creation Monitor image generation progress (12 images) Review HTML moodboard generation Confirm PDF creation Check email delivery

Step 4: Configure Form Embedding (Optional)

Embed the form on your website:

<iframe src="YOUR_N8N_FORM_WEBHOOK_URL" width="100%" height="800" frameborder="0"> </iframe>

🔄 Workflow Structure

1. Form Input & Data Extraction Moodboard Form** - Collects project title, description, and user email Email Extractor** - Extracts username from email for folder organization

2. Storage Setup Create a Folder** - Creates personalized Nextcloud directory using email username

3. AI Concept Generation Conceptualization Agent** (Claude Sonnet 4) - Analyzes design brief and generates 12 detailed image prompts (300-500 words each) Images 1-11: Individual design elements (furniture, materials, details, styling) Image 12: Comprehensive 3D rendered view integrating all elements

4. Image Processing Loop Concept Splitter** - Separates 12 prompts into individual items Loop Over Items** - Processes each prompt sequentially: Image Generator - Sends prompt to Hugging Face FLUX.1-schnell API Upload Image - Stores generated image in Nextcloud folder Share a File - Creates public shareable URL Set Image Title and URL - Formats data for aggregation

5. URL Collection URL Aggregate** - Combines all 12 public image URLs Clean URLs** - Extracts and formats URLs into a structured list

6. Moodboard Compilation Moodboard Generator Agent** (Google Gemini 2.5 Pro) - Creates professional two-page HTML document: Page 1: Visual moodboard with all 12 images (Image #12 prominently featured 2-3x larger) Page 2: Administrative summary with design overview, color palette, materials, and project details

7. PDF Generation & Delivery Binary Converter** - Transforms HTML to base64-encoded binary format PDF Creator** - Converts HTML to print-ready PDF via Gotenberg service Send PDF** - Emails final moodboard PDF to user

🎯 Node Descriptions

Moodboard Form Collects moodboard generation requests for any design topic. Users input a title, detailed description (colors, materials, patterns, textures, lighting), and email address for delivery.

Email Extractor Extracts the username portion from email addresses, optionally stripping "+tags" for clean folder naming and user identification.

Create a Folder Creates a dedicated Nextcloud folder using the extracted email username, organizing moodboard outputs by user.

Conceptualization Agent AI agent that analyzes design briefs to generate 12 detailed image prompts (300-500 words each). Performs conceptual analysis of styles, colors, materials, and spatial requirements, outputting structured JSON.

Concept Splitter Splits the 12 generated image prompts into individual items for parallel processing through the image generation pipeline.

Loop Over Items Processes each prompt sequentially, generating images, uploading to Nextcloud, and creating public URLs.

Image Generator Sends detailed prompts to Hugging Face FLUX.1-schnell API for AI-powered image generation, transforming written design concepts into high-quality visuals.

Upload Image Uploads each generated moodboard image to the user's Nextcloud folder with appropriate naming conventions.

Share a File Creates publicly shareable Nextcloud links for each uploaded image, enabling external viewing without authentication.

Set Image Title and URL Formats image metadata (title and URL) for downstream aggregation.

URL Aggregate Combines all 12 image URLs into a single consolidated output for moodboard compilation.

Clean URLs Extracts and formats URLs from the aggregated data into a clean, structured list with count.

Moodboard Generator Agent Transforms design concepts into professional two-page HTML moodboards. Analyzes project details and 12 image URLs, selecting appropriate visual styles. Creates artistic Page 1 with Image #12 as hero element, and comprehensive Page 2 with design documentation.

Binary Converter Prepares HTML for PDF conversion by transforming it into binary format with proper encoding and filename ("index.html") for Gotenberg compatibility.

PDF Creator Converts HTML moodboard into print-ready PDF with proper A4 dimensions, page breaks, and high-quality image resolution.

Send PDF Emails the finalized PDF moodboard to the user's submitted email address with project details and PDF attachment.

🎨 Customization Options

Design Styles The Moodboard Generator Agent automatically selects from 10 layout styles: Modern Sectional Grid Material Board Collage Editorial Magazine Clean Minimalist Split-Screen Dramatic Asymmetric Feature Centered Showcase Modular Block System Organic Flow Layered Depth

Image Generation Models Replace FLUX.1-schnell with alternatives in the "Image Generator" node: black-forest-labs/FLUX.1-dev - Higher quality, slower stabilityai/stable-diffusion-xl-base-1.0 - Classic SD XL

Email Templates Customize the email message in the "Send PDF" node to include: Brand messaging Next steps Support contact information Pricing information

🐛 Troubleshooting

Images Not Generating Verify Hugging Face API token is valid Check API rate limits and quotas Increase timeout in "Image Generator" node (default: unlimited)

PDF Generation Fails Ensure Gotenberg service is accessible Verify HTML output contains all 12 image URLs Check timeout settings (default: 360s) Review Gotenberg logs for specific errors

Nextcloud Upload Errors Confirm folder creation succeeded Verify Nextcloud credentials and permissions Check available storage space Ensure WebDAV is enabled

Email Not Received Verify Gmail OAuth2 credentials Check spam/junk folders Confirm email address is valid Review Gmail API quotas

📊 Performance Notes

Average execution time**: 5-8 minutes (depends on image generation) Image generation**: ~20-30 seconds per image (12 images = 4-6 minutes) PDF generation**: ~30-60 seconds Total data processed**: ~15-25 MB per workflow execution

🔐 Security Considerations

Store all API keys in n8n credentials (never hardcode) Use environment variables for sensitive configuration Implement rate limiting on the form webhook Consider adding CAPTCHA to prevent abuse Regularly rotate API keys and passwords Use HTTPS for all external communications

📝 License & Attribution

This template is provided as-is for the n8n community. Feel free to modify and adapt to your needs.

AI Models Used: Claude Sonnet 4 (Anthropic via OpenRouter) Google Gemini 2.5 Pro (Google) FLUX.1-schnell (Black Forest Labs via Hugging Face)

🤝 Support & Contributions

For questions or improvements, please reach out through the n8n community forum or submit issues/PRs to enhance this template.

Created by: Jameson Kanakulya
Template Version: 1.0
Last Updated: November 2025