Previously, the maximum output tokens for image descriptions was set to 256.
This was often too restrictive, leading to truncated or incomplete descriptions.
Increasing the limit to 1024 allows for more comprehensive and detailed image
descriptions while still encouraging conciseness.