feat/<short-name> either add parameter "extract_image_block_output_dir" in unstructured api call of shared.PartitionParameters or include base46 encoded images in meta data when chunking stratergy ="by_title"

[mahendra867]

Is your feature request related to a problem? Please describe.
. Ex. I'm always frustrated when Yes. I'm currently using the Unstructured API with chunking_strategy="by_title" to segment documents for downstream semantic search and RAG applications. However, when using this chunking strategy, I’m unable to retrieve the Base64-encoded images in the document (i.e., no image_base64 in the metadata of returned elements). Additionally, the parameter "extract_image_block_output_dir"—which allows extracted image blocks to be saved—is unavailable or not functioning when is used. unstructured by api call

This prevents me from building a robust multimodal pipeline that includes both structured text chunks and their corresponding visual content.

Describe the solution you'd like
I’d like one of the following to be supported when using chunking_strategy="by_title" in the Unstructured API:

(Preferred) Embed Base64-encoded image (image_base64) in the metadata of returned elements—just as is supported in r hi-res strategies.

(Alternative) Enable the extract_image_block_output_dir parameter to work unstructured by api call , so that extracted image blocks are saved to a specified directory even during chunked extraction.

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.