Want Perfect AI Saree Photos? 5 Things You Should Never Do With Nano Banana
The viral saree trend can fail if you use wrong prompts. Check what never to do while generating AI saree portraits with Google Gemini Nano Banana.
image for illustrative purpose
A new wave of artificial intelligence portraits has gripped social media, where users are reshaping their photos into Bollywood-era saree looks through Google’s Gemini Nano Banana tool.
The model, launched by DeepMind in August 2025, is designed for AI-powered image generation and editing. Its latest use case — the saree transformation trend — is generating huge traction online. People are sharing stylized images that mimic the look of classic film posters, often with warm-toned lighting, vintage backdrops, and floral hairstyles.
Instagram, Facebook, and similar platforms are crowded with such edits, drawing participation from ordinary users as well as influencers. The craze follows earlier viral experiments such as ChatGPT’s Ghibli-inspired images.
Still, not everyone is pleased with the results. Many complained that the portraits sometimes distort their appearance, with mismatched features or unrealistic face structures. According to AI experts, the problem often arises from how prompts are framed.
What not to do?
Users experimenting with the tool have been cautioned against uploading blurry photographs or large group shots, as the model struggles to map faces clearly. Short and precise prompts, they say, are more effective than long descriptive ones. Directions that include the saree’s fabric or color, the lighting style, and the kind of background also improve the final outcome.
Facial consistency is another challenge. To reduce changes in appearance, some suggest adding instructions such as “keep the same features” or “do not modify face shape.” Similarly, specifying the backdrop avoids awkward results and ensures the final portrait resembles a cinematic still.
