some pptx,docx,pdf documents might have images embeded inside. Could we discuss about the implementation of parsing the image via vision model to extract to text format and index into vector database. In this way, it will provide better answer when user ask question about the images.