Descriptions:
This video tests OpenAI’s GPT Image 1.5 against Google’s Imagen 3 (marketed here as Nano Banana Pro) for AI-generated UGC advertisement creation, running both models in parallel through an identical n8n automation workflow. The benchmark framing comes from LM Arena’s image editing leaderboard, where GPT Image 1.5 scores 1409 against Imagen 3’s 1406—a margin close enough that hands-on results could go either way.
Zubair Trabzada builds and walks through a complete n8n pipeline: a form accepts a product photo and text description, an OpenAI vision model analyzes the image and generates an optimized prompt, and both image models run simultaneously. The resulting images are then passed to Veo 3.1 to animate into short UGC-style video ads. The test product is a Chanel perfume, with both models producing polished ad creatives featuring a styled male model and strong product emphasis—with subtle differences in background blur, lighting placement, and subject framing.
The practical takeaway is that GPT Image 1.5 and Imagen 3 perform at near-identical quality for product ad editing workflows, and model choice may come down to API pricing, existing integration setup, or stylistic preference. The full n8n workflow JSON, prompting guides, and all resources are made available free through the AI Workshop community.
📺 Source: Zubair Trabzada | AI Workshop · Published December 18, 2025
🏷️ Format: Comparison
