Tried to tweak the prompt for closer comparison with "it should be beautiful, close up and golden hour"
@fofrai
-
GPT Image 2 Generates Hidden Word in Pride Flag Design
By
–
Not sure why this output is so dark (tried a few times), but here's gpt-image-2's version of this prompt: > make a photo with a hidden word embedded into it, make the word hard to see until you spot it, then you can't unsee it. use the word "fofr". form the word out of a pride
-
User critique chat-based image generation interfaces
By
–
I don't like any image generation with a chat UI, especially if you're waiting for one image at a time.
-
Nano Banana Pro outperforms gpt-image-2 in satellite map accuracy
By
–
> a satellite map of London, showing a close up of the area around Big Ben Left: Nano Banana Pro
Right: gpt-image-2 Seeing NBP has more accuracy than gpt-image-2 for this prompt. Westminster bridge should be next to Big Ben. The road layouts, park positions and buildings are -
GPT-Image-2 Demonstrates Advanced Screenshot Generation Capabilities
By
–
gpt-image-2 is pretty good. > show me a screenshot of a mac desktop, large terminal window visible, doing something in the terminal with an expressive TUI layout related to a world sim
-
Accessibility concerns: hearing impaired users and language barriers
By
–
That's because you think everyone with a hearing impairment can read English.
-
AI Still Far From Accomplishing This Task
By
–
FWIW, we're still so far away from AI being able to do this.https://t.co/GGTpf8AYhL
— fofr (@fofrAI) 21 avril 2026FWIW, we're still so far away from AI being able to do this.
-
AI Animation Accessibility: Studios Should Fund Animator-Created Versions
By
–
I would love for studios to pay animators to make these accessible versions instead. Also, AI will work with live action performances too.