📝 Summary
A GPT-Image-2 community comparison case using "Model comparison: GPT Image 2 vs Nano Banana 2" to review image quality, layout control, and prompt-following across models.
🎨 Chinese Prompt
原來ChatGPT嘅圖像生成Model又更新咗
講多無謂! 直接用同樣嘅Prompt同參考圖測試下
(左) GPT Image 2 (右) Nano Banana 2
本身GPT Image都有主體同背景光線強度不一嘅問題
而家仲要整到個主體嘅頭,身,腳嘅比例好奇怪😮
我會用災難級嚟形容GPT Image 2嘅表現
🎨 English Prompt
It turns out the ChatGPT image generation model has been updated again
No more talking! Let’s test it directly using the same prompt and reference image
(Left) GPT Image 2 (Right) Nano Banana 2
Originally, GPT Image already had issues with inconsistent subject and background lighting intensity
Now it’s even worse—the proportions of the subject’s head, body, and legs are oddly strange 😮
I would describe GPT Image 2’s performance as catastrophic
📷 Images
