Embedding Images

[uwilken]

Hi,
I am setting up a RAG application next to a CMS and beside text I also plan to embed image data to also cover images via similarity/semantic search.
While feeding the data into a vector store, I use a chatClient to send along image data and let the AI describe the image and add the returned description to the vector db.
I lately ran some tests with the Transformers (ONNX) Embeddings which works quite well. Is there a way to also embed image descriptions this way directly without asking the chat client for an image description in advance.

Has anyone managed to do this?
All tips are very welcome.