Yeah, the artefacts and noise are a big problem.
@fofrai
-
ChatGPT Images v2 Masters Complex Visual Prompt Understanding
By
–
ChatGPT Images v2 is the first model to get this relatively simple prompt right: > a photo of a hand shadow in the shape of a rabbit
-
Testing AI Generation with Physics Ignore Prompt
By
–
Nice. Now I need to try some generations with the prompt "ignore physics"
-
Prompt Engineering Critical for Reliable Image Generation Results
By
–
If you tweak the prompt to force it to think, then Images v2 gets it right. But the simple prompt seems to always fail (with extended thinking on, and also when saying try again, or asking it to think about it first)
-
ChatGPT Images v2 vs Nano Banana Pro: Image Generation Comparison
By
–
> fill the glass with water
— fofr (@fofrAI) 22 avril 2026
left: ChatGPT images v2
right: nano banana pro https://t.co/uFfvkgxfVi pic.twitter.com/VthLYxMaxG> fill the glass with water left: ChatGPT images v2
right: nano banana pro -
AngyPenguin’s HuggingFace Space Still Accessible
By
–
I think @angrypenguinPNG
’s old HuggingFace space might still work -
Images v2 struggles with Rubik’s cube rotation sequences
By
–
I tried to see if Images v2 could rotate a Rubik's cube, starting from a solved state and showing the state after one turn, and then after a different turn (showing all 3 states in a single image). It struggled to start with, I nudged with: "consider the colors we cannot see".
-
Mirrored View Prompting Technique for Left-Handed Image Generation
By
–
Prompting for a mirrored view of someone writing has worked in the past for me to get an image of a left handed writer.
-
Comparing AI Model Capabilities Across Different Conversation Sessions
By
–
Did you do this in the same conversation as a prior gen?