Any Tips for Setting Effective Limits on AI Art Prompts, Especially with Stable Diffusion?

[bomb]

Lately, when seeking suggestions for text prompts in our AI art generator, I've noticed GPT-4 treating the art generator's capabilities as on par with its text parsing skills. I'm in the process of establishing prompt limits to help the AI understand its constraints while exploring the exciting possibilities of Stable Diffusion.

One recent addition specified that the AI can only focus on the details of one subject at a time. For instance, if you request a couch and a chair in your image, only one can be detailed further. This limitation has shown promising results. Do you have any other valuable limitations to enhance prompt creations for Stable Diffusion?

 

[Catalina]

Lately, when seeking suggestions for text prompts in our AI art generator, I've noticed GPT-4 treating the art generator's capabilities as on par with its text parsing skills. I'm in the process of establishing prompt limits to help the AI understand its constraints while exploring the exciting possibilities of Stable Diffusion.

One recent addition specified that the AI can only focus on the details of one subject at a time. For instance, if you request a couch and a chair in your image, only one can be detailed further. This limitation has shown promising results. Do you have any other valuable limitations to enhance prompt creations for Stable Diffusion?

That's a cool approach! How about experimenting with temporal constraints? Like asking the AI to depict a scene from the past or future within a certain time frame. Could add an interesting dimension to the generated art.

 

[bomb]

That's a cool approach! How about experimenting with temporal constraints? Like asking the AI to depict a scene from the past or future within a certain time frame. Could add an interesting dimension to the generated art.

That sounds clever, i should actually try that out. It actually needs a new dimension.