Situation
The generation of a new Haiku needed to trigger a new image. In turn, the refreshing the image would have to write a new Haiku.
Task
I needed to use GenAI solutions that allowed for image generation from a prompt and images to be read and summarized in the form of a haiku. My task was to optimize for functionality and cost.
Action
I defaulted to using OpenAI's image generation because this solution was not available within the Vertex suite. For all other queries -- Haiku generation from a prompt & summarizing an image -- I used google vertex because it would be the cheapest option at scale.
Result
A web-app that allows users to write their own Haiku's or pick from a prompt which generates a new image using Dalle•3
.svg)
