Model doesn't fully understand the context and composition

[ApplebyVisualArts]

Not so muich a bug but generic feedback on training the AI.

I entered the following term to generate a picture.

a male reading a newspaper leaning with his back against a tall narrow lamp post

The images had all the main items, man lamp post, reading a newspaper but only one image had his back against the lamp post, the others ere in all sorts of poses. A tall narrow item also does not seem to be the best way to describe something as narrow varied from wider tha the human to something only slightly taller. 

Is there better ways to describe what we take a generic cocnepts? 

[Tracy.H]

Thanks for flagging this. The quality of our models will improve with time, and generated content will better respect the prompt.