Whats up with this comically bad AI picture? Almost every model i know can do way better.

I am going to assume that the author was visually impaired themselves because otherwise why not try to generate again to get sm that at least vagely makes sense.

I fed the picture to chatgpt and asked "Analyze the style and all elements in this picture. Use DallE to recreate it "

Not exactly great art either but like at least the bench exists properly

Thanks for posting this! Unfortunately we haven’t had any feedback yet - perhaps that’s an indication that the apps are not sufficiently accessible at the moment, which means they can’t really be used with a screen reader. But we hope to improve this