Closed moodler closed 1 year ago
Image -> Text API got: "a large dog wearing a turban with a dog behind it in the mountain" for this image
I've improved it for most cases, but the image to text tech isn't great.
It is sometimes misinterpreting the images, but I don't think there is much we can do about that.
Edubot is now informed of both it's own interpretation of the image, as well as the prompt that generated the image. So it will prefer the prompt info and only use it's own interpretation for details not included in the prompt