Did ChatGPT Just Kill Nano Banana?

no they didnt

In the example where you told it to change to snowing outside, you failed to mention an important consideration for your prompt. You told ChatGPT to change the image to a "cozy winter morning". You were saying ChatGPT wasn't adhering to your prompt because it changed elements on the table, but what it changed on the table was to switch the meal from a Turkey dinner to breakfast with coffee to match the request for a "cozy morning". That is actually very smart of the model to understand you don't usually eat turkey and drink wine for breakfast and change it accordingly!

As the co-founder of Deepmind just said the n Moonshots... It's not an AI race. There's no destination. They'll be constantly 1-Uping each other until we can't tell the difference or there's a unimodel.

OpenAI has too many stupid guidelines. So I am not excited. I asked it to generate an image of a girl reading a book, zoom in on her hands holding the book. It said it could not because it's a minor. Like WTF. Nano banana did it with no issues. OpenAI really need to chill.🤣

The amount of censorship just totally kills it for me :))

I noticed your first prompt in the multi-step editing and consistency exercise included the word "morning" so for me GPT changing the elements of the scene that don't read as a "morning" scene, like wine and turkey, is it actually demonstrating better prompt adherence than Nano Banana did there 🤷‍♀️...at least until the little girl became immortalized as a 5 year old in subsequent turns 😆

I just tested it. It looks promising. Battle is ongoing.

Nano Banana Pro STILL the undisputed champion lol

I really like the prebuilt presets, especially the sketch one.

Just run a test and created an infographic about a specific youtube channel. It gave a great result on the first try with not a single spelling error - even the discord url links were correct on every single letter. This may be my favorite to-go image tool for now.