Generate Ad Image Variations Using GPT-4 and Dumpling AI

This n8n workflow automates the process of creating multiple creative variations of an advertisement image using AI tools. It begins with a user submitting brand info and an original ad image through a form trigger. The image is then uploaded to Google Drive for analysis, where GPT-4 describes its visual style. Concurrently, the workflow analyzes the brand’s website to understand its visual identity. A LangChain agent combines these insights to generate 10 subtle variation prompts aimed at preserving the core subject while altering background, lighting, mood, and color. Each prompt is sent to Dumpling AI to produce new image variations, which are then stored in Google Drive and their URLs logged into Google Sheets for tracking. This workflow is ideal for advertisers and marketers seeking quick, consistent testing of ad creative variations without manual graphic design.