In this video, I show you how to build a real web app/SaaS that users can log into, hit “generate,” and create UGC-style videos using just a prompt + an image. The whole idea is simple: Google dropped VEO 3.1 (best video model right now) and Nano Banana (best image model right now, with strong quality + character consistency), so I combine both into one workflow. For the frontend, I use Lovable to generate a working dashboard just by chatting, and for the backend I use n8n to receive requests, run the AI steps, and return results back to the app. I start by getting n8n running the cheapest way: self-hosting it on a Hostinger VPS so you unlock unlimited workflows and concurrent executions. Then I build two webhook-based flows in n8n: one for “image” and one for “video,” and I wire them into Lovable via POST requests. For the AI generation, I use fal.ai because it supports Nano Banana + VEO 3.1 and charges pay-as-you-go (no subscription). Finally, I show how to handle uploaded images (binary), upload them to Google Drive to get a URL, call the Nano Banana edit endpoint, poll for completion with a wait + status check loop, and respond back to Lovable with a clean JSON payload so the user can see/download the result.