Aleena Amir Profile picture
Turning thoughts into art.
Apr 22 8 tweets 2 min read
🧵 I tested OpenAI’s GPT Image 2 model in ChatGPT with multi-image prompts

Instead of one image, I asked for full sets (brand kits, UI systems, cinematic scenes, and more).

Wanted to see if it stays consistent across images.

Prompts are in ALT.

1. Worldbuilding Set Create a complete visual worldbuilding set for a futuristic desert civilization powered by solar technology, multiple images including architecture, characters, clothing, vehicles, and maps, cohesive design language, cinematic realism, ultra detailed. 2. Consistency Stress Test Create a highly consistent multi-image set where the same young woman with short black hair, freckles, and a red jacket appears in different environments (city, beach, mountains, night street), maintaining exact facial features, proportions, and outfit details, ultra-realistic, cinematic lighting variations.