Support for AI Vision Vectorizer (Multimodal Embedding) when using Integrated Vectorization

[xxxpsyduck]

Azure AI Search supports creating multimodal embedding using AI Vision Vectorizer now:

This will create an unified embedding field in the index with dimension 1024:

I already set the following env vars (some are censored):

AZURE_DOCUMENTINTELLIGENCE_SERVICE="doc..."
AZURE_ENV_NAME="azure-search-openai-demo"
AZURE_LOCATION="East US"
AZURE_OPENAI_API_VERSION="2025-04-01-preview"
AZURE_OPENAI_CHATGPT_DEPLOYMENT="model-router"
AZURE_OPENAI_CHATGPT_DEPLOYMENT_SKU="GlobalStandard"
AZURE_OPENAI_CHATGPT_DEPLOYMENT_VERSION="2025-08-07"
AZURE_OPENAI_CHATGPT_MODEL="model-router"
AZURE_OPENAI_LOCATION="East US 2"
AZURE_OPENAI_RESOURCE_GROUP="...res"
AZURE_OPENAI_SERVICE="...eastus2"
AZURE_RESOURCE_GROUP="...res"
AZURE_SEARCH_FIELD_NAME_EMBEDDING="content_embedding"
AZURE_SEARCH_INDEX="multimodal-rag-test"
AZURE_SEARCH_SEMANTIC_RANKER="disabled"
AZURE_SEARCH_SERVICE="...-ai-search-service"
AZURE_SEARCH_SERVICE_LOCATION="East US"
AZURE_SEARCH_SERVICE_RESOURCE_GROUP="...res"
AZURE_STORAGE_ACCOUNT="..."
AZURE_STORAGE_CONTAINER="...files"
AZURE_SUBSCRIPTION_ID="..."
RAG_SEARCH_TEXT_EMBEDDINGS="false"
USE_FEATURE_INT_VECTORIZATION="true"
USE_MULTIMODAL="true"

May I know how I could my existing index above with current setup?