Merge pull request #4222 from HeidiSteen/heidist-freshness

prmerger-automator[bot] · web-flow · commit 09aab240aad5 · 2025-04-18T19:44:53.000Z
AI Vision multimodal doc updates
diff --git a/articles/search/cognitive-search-defining-skillset.md b/articles/search/cognitive-search-defining-skillset.md
@@ -8,7 +8,7 @@ ms.service: azure-ai-search
 ms.custom:
   - ignite-2023
 ms.topic: conceptual
-ms.date: 12/06/2024
+ms.date: 04/18/2025
 ---
 
 # Create a skillset in Azure AI Search
@@ -192,7 +192,7 @@ Skills read from and write to an enriched document. Skill inputs specify the ori
   | `source`: `/document/some-named-field` | For text-based skills, such as entity recognition or key phrase extraction, the origin should be a field that contains sufficient text to be analyzed, such as a *description* or *summary*. |
   | `source`: `/document/normalized_images/*` | For image content, the source is image that's been normalized during document cracking. |
 
-If the skill iterates over an array, both context and input source should include `/*` in the correct positions.
+If the skill iterates over an array, both context and input source should include `/*` in the correct positions. For more information about the complete syntax, see [Skill context and input annotation language](cognitive-search-skill-annotation-language.md).
 
 ## Define outputs
 
diff --git a/articles/search/cognitive-search-skill-vision-vectorize.md b/articles/search/cognitive-search-skill-vision-vectorize.md
@@ -9,44 +9,55 @@ ms.custom:
   - build-2024
   - references_regions
 ms.topic: reference
-ms.date: 08/05/2024
+ms.date: 04/18/2025
 ---
 
 # Azure AI Vision multimodal embeddings skill
 
 > [!IMPORTANT] 
-> This skill is in public preview under [Supplemental Terms of Use](https://azure.microsoft.com/support/legal/preview-supplemental-terms/). The [2024-05-01-Preview REST API](/rest/api/searchservice/skillsets/create-or-update?view=rest-searchservice-2024-05-01-Preview&preserve-view=true) supports this feature.
+> This skill is in public preview under [Supplemental Terms of Use](https://azure.microsoft.com/support/legal/preview-supplemental-terms/). The [2024-05-01-Preview REST API](/rest/api/searchservice/skillsets/create-or-update?view=rest-searchservice-2024-05-01-Preview&preserve-view=true) and newer preview APIs support this feature.
 
 The **Azure AI Vision multimodal embeddings** skill uses Azure AI Vision's [multimodal embeddings API](/azure/ai-services/computer-vision/concept-image-retrieval) to generate embeddings for image or text input.
 
-The skill is only supported in search services located in a region that supports the [Azure AI Vision Multimodal embeddings API](/azure/ai-services/computer-vision/how-to/image-retrieval). Review [region availability for multimodal embeddings](/azure/ai-services/computer-vision/overview-image-analysis?tabs=4-0#region-availability). Your data is processed in the [Geo](https://azure.microsoft.com/explore/global-infrastructure/data-residency/) where your model is deployed. 
+This skill must be [attached to a billable Azure AI multi-service resource](cognitive-search-attach-cognitive-services.md) for transactions that exceed 20 documents per indexer per day. Execution of built-in skills is charged at the existing [Azure AI services pay-as-you go price](https://azure.microsoft.com/pricing/details/cognitive-services/). 
 
-> [!NOTE]
-> This skill is bound to Azure AI services and requires [a billable resource](cognitive-search-attach-cognitive-services.md) for transactions that exceed 20 documents per indexer per day. Execution of built-in skills is charged at the existing [Azure AI services pay-as-you go price](https://azure.microsoft.com/pricing/details/cognitive-services/).
->
-> In addition, image extraction is [billable by Azure AI Search](https://azure.microsoft.com/pricing/details/search/).
->
+In addition, image extraction is [billable by Azure AI Search](https://azure.microsoft.com/pricing/details/search/).
+
+Location of resources is an important consideration. Because you're using a preview API version to create a skillset that contains preview skills, you have the option of a [keyless connection](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection), which relaxes the region requirement. However, if you're connecting with an API key, then Azure AI Search and Azure AI multi-service must be in the same region.
+
++ First, find a [supported region for multimodal embeddings](/azure/ai-services/computer-vision/overview-image-analysis?tabs=4-0#region-availability).
+
++ Second, verify the [region provides AI enrichment](search-region-support.md).
+
+The Azure AI multi-service resource is used for billing purposes only. Content processing occurs on separate resources managed and maintained by Azure AI Search within the same geo. Your data is processed in the [Geo](https://azure.microsoft.com/explore/global-infrastructure/data-residency/) where your resource is deployed. 
 
 ## @odata.type  
 
 Microsoft.Skills.Vision.VectorizeSkill
 
 ## Data limits
 
-The input limits for the skill can be found in [the Azure AI Vision documentation](/azure/ai-services/computer-vision/concept-image-retrieval#input-requirements) for images and text respectively. Consider using the [Text Split skill](cognitive-search-skill-textsplit.md) if you need data chunking for text inputs.
+The input limits for the skill can be found in the [Azure AI Vision documentation](/azure/ai-services/computer-vision/concept-image-retrieval#input-requirements) for images and text respectively. Consider using the [Text Split skill](cognitive-search-skill-textsplit.md) if you need data chunking for text inputs. 
+
+Applicable inputs include:
+
++ Image input file size must be less than 20 megabytes (MB). Image size must be greater than 10 x 10 pixels and less than 16,000 x 16,000 pixels.
++ Text input string must be between (inclusive) one word and 70 words.
 
 ## Skill parameters
 
 Parameters are case-sensitive.
 
 | Inputs | Description |
 |---------------------|-------------|
-| `modelVersion` | (Required) The model version to be passed to the Azure AI Vision multimodal embeddings API for generating embeddings. It's important that all embeddings stored in a given index field are generated using the same `modelVersion`. For information about version support for this model, refer to [multimodal embeddings](/azure/ai-services/computer-vision/concept-image-retrieval#what-are-vector-embeddings).|
+| `modelVersion` | (Required) The model version (`2023-04-15`) to be passed to the Azure AI Vision multimodal embeddings API for generating embeddings. Vector embeddings can only be compared and matched if they're from the same model type. Images vectorized by one model won't be searchable through a different model. The latest Image Analysis API offers two models, version `2023-04-15` which supports text search in many languages, and the legacy `2022-04-11` model which supports only English. Azure AI Search uses the newer version. |
 
 ## Skill inputs
 
+Skill definition inputs include name, source, and inputs. The following table provides valid values for name of the input. You can also specify recursive inputs. For more information, see the [REST API reference](/rest/api/searchservice/skillsets/create?view=rest-searchservice-2025-03-01-preview#inputfieldmappingentry&preserve-view=true) and [Create a skillset](cognitive-search-defining-skillset.md).
+
 | Input	 | Description |
-|--------------------|-------------|
+|--------|-------------|
 | `text` | The input text to be vectorized. If you're using data chunking, the source might be `/document/pages/*`. |
 | `image` | Complex Type. Currently only works with "/document/normalized_images" field, produced by the Azure blob indexer when ```imageAction``` is set to a value other than ```none```. |
 | `url` | The URL to download the image to be vectorized. |
@@ -62,15 +73,15 @@ Only one of `text`, `image` or `url`/`queryString` can be configured for a singl
 
 ## Sample definition
 
-For text input, consider a record that has the following fields:
+For text input, consider a blob that has the following content:
 
 ```json
 {
-    "content": "Microsoft released Windows 10."
+    "content": "Forests, grasslands, deserts, and mountains are all part of the Patagonian landscape that spans more than a million square  kilometers of South America."
 }
 ```
 
-Then your skill definition might look like this:
+For text inputs, your skill definition might look like this:
 
 ```json
 { 
@@ -85,14 +96,15 @@ Then your skill definition might look like this:
     ], 
     "outputs": [ 
         { 
-            "name": "vector"
+            "name": "vector",
+            "targetName": "text_vector"
         } 
     ] 
 } 
 
 ```
 
-For image input, your skill definition might look like this:
+For image input, a second skill definition in the same skillset might look like this:
 
 ```json
 {
@@ -107,13 +119,14 @@ For image input, your skill definition might look like this:
     ],
     "outputs": [
         {
-            "name": "vector"
+            "name": "vector",
+            "targetName": "image_vector"
         }
     ]
 }
 ```
 
-If you want to vectorize images directly from your blob storage datasource, your skill definition might look like this:
+If you want to vectorize images directly from your blob storage data source rather than extract images during indexing, your skill definition should specify a URL, and perhaps a SAS token depending on storage security. For this scenario, your skill definition  might look like this:
 
 ```json
 {
@@ -132,19 +145,20 @@ If you want to vectorize images directly from your blob storage datasource, your
     ],
     "outputs": [
         {
-            "name": "vector"
+            "name": "vector",
+            "targetName": "image_vector"
         }
     ]
 }
 ```
 
 ## Sample output
 
-For the given input text, a vectorized embedding output is produced.
+For the given input, a vectorized embedding output is produced. Output is 1,024 dimensions, which is the number of dimensions supported by the Azure AI Vision multimodal API.
 
 ```json
 {
-  "vector": [
+  "text_vector": [
         0.018990106880664825,
         -0.0073809814639389515,
         .... 
@@ -153,7 +167,7 @@ For the given input text, a vectorized embedding output is produced.
 }
 ```
 
-The output resides in memory. To send this output to a field in the search index, you must define an [outputFieldMapping](cognitive-search-output-field-mapping.md) that maps the vectorized embedding output (which is an array) to a [vector field](vector-search-how-to-create-index.md). Assuming the skill output resides in the document's **vector** node, and **content_vector** is the field in the search index, the outputFieldMapping in indexer should look like:
+The output resides in memory. To send this output to a field in the search index, you must define an [outputFieldMapping](cognitive-search-output-field-mapping.md) that maps the vectorized embedding output (which is an array) to a [vector field](vector-search-how-to-create-index.md). Assuming the skill output resides in the document's **vector** node, and **content_vector** is the field in the search index, the outputFieldMapping in the indexer should look like:
 
 ```json
   "outputFieldMappings": [
@@ -164,7 +178,7 @@ The output resides in memory. To send this output to a field in the search index
   ]
 ```
 
-For mapping image embeddings to the index, you'll need to use the [Index Projections](index-projections-concept-intro.md) feature. The payload for `indexProjections` might look something like this:
+For mapping image embeddings to the index, you use [index projections](index-projections-concept-intro.md). The payload for `indexProjections` might look something like the following example. image_content_vector is a field in the index, and it's populated with the content found in the **vector** of the **normalized_images** array.
 
 ```json
 "indexProjections": {
@@ -175,7 +189,7 @@ For mapping image embeddings to the index, you'll need to use the [Index Project
             "sourceContext": "/document/normalized_images/*",
             "mappings": [
                 {
-                    "name": "content_vector",
+                    "name": "image_content_vector",
                     "source": "/document/normalized_images/*/vector"
                 }
             ]
diff --git a/articles/search/search-region-support.md b/articles/search/search-region-support.md
@@ -27,7 +27,7 @@ Some features take a dependency on other Azure services or infrastructure that a
 | [Availability zones](search-reliability.md#availability-zone-support) | Divides a region's data centers into distinct physical location groups, providing high availability within the same geo. | Regional support is noted in this article. |
 | [Semantic ranker](semantic-search-overview.md) | Takes a dependency on Microsoft-hosted models in specific regions. | Regional support is noted in this article. |
 | [Query rewrite](semantic-how-to-query-rewrite.md) | Takes a dependency on Microsoft-hosted models in specific regions. | Regional support is noted in this article. |
-| [AI service integration](cognitive-search-concept-intro.md) | Refers to [built-in skills](cognitive-search-predefined-skills.md) that make internal calls to Azure AI for enrichment and transformation during indexing. Integration requires that Azure AI Search coexists with an [Azure AI services multi-service account](/azure/ai-services/multi-service-resource#azure-ai-services-resource-for-azure-ai-search-skills) in the same physical region. You can bypass region requirements if you use [identity-based connections](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection), currently in public preview. | Regional support is noted in this article. |
+| [AI enrichment](cognitive-search-concept-intro.md) | Refers to [built-in skills](cognitive-search-predefined-skills.md) that make internal calls to Azure AI for enrichment and transformation during indexing. Integration requires that Azure AI Search coexists with an [Azure AI services multi-service account](/azure/ai-services/multi-service-resource#azure-ai-services-resource-for-azure-ai-search-skills) in the same physical region. You can bypass region requirements if you use [identity-based connections](cognitive-search-attach-cognitive-services.md#bill-through-a-keyless-connection), currently in public preview. | Regional support is noted in this article. |
 | [Azure OpenAI integration](vector-search-integrated-vectorization.md)  | Refers to the AzureOpenAIEmbedding skill and vectorizer that make internal calls to deployed embedding models on Azure OpenAI. | Check [Azure OpenAI model region availability](/azure/ai-services/openai/concepts/models#model-summary-table-and-region-availability) for the most current list of regions for each embedding and chat model. Specific Azure OpenAI models are in fewer regions, so check for model availability first, and then verify Azure AI Search is available in the same region.|
 | [Azure AI Foundry integration](vector-search-integrated-vectorization-ai-studio.md) | Refers to skills and vectorizers that make internal calls to the models hosted in the model catalog. | Check [Azure AI Foundry region availability](/azure/ai-foundry/reference/region-support) for the most current list of regions. |
 | [Azure AI Vision 4.0 multimodal APIs](search-get-started-portal-image-search.md) | Refers to the Azure AI Vision multimodal embeddings skill and vectorizer that call the multimodal embedding API. | Check the [Azure AI Vision region list](/azure/ai-services/computer-vision/overview-image-analysis#region-availability) first, and then verify Azure AI Search is available in the same region.|
@@ -36,11 +36,11 @@ Some features take a dependency on other Azure services or infrastructure that a
 
 You can create an Azure AI Search resource in any of the following Azure public regions. Almost all of these regions support [higher capacity tiers](search-limits-quotas-capacity.md#service-limits). Exceptions are noted where they apply.
 
-AI service integration refers to internal connections to an Azure AI services multi-service account and doesn't include Azure OpenAI integration.
+AI enrichment refers to internal connections to an Azure AI services multi-service account and doesn't include Azure OpenAI integration.
 
 ### Americas
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | Brazil South​​ ​| ✅ |  | ✅ |  |
 | Canada Central​​ | ✅ | ✅ | ✅ |  |
@@ -58,7 +58,7 @@ AI service integration refers to internal connections to an Azure AI services mu
 
 ### Europe
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | North Europe​ <sup>1</sup>​ | ✅ | ✅ | ✅ | ✅ |
 | West Europe​​ | ✅ | ✅ | ✅ |  |
@@ -80,7 +80,7 @@ AI service integration refers to internal connections to an Azure AI services mu
 
 ### Middle East
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | Israel Central​ <sup>1</sup> |  | ✅ |  |  |
 | Qatar Central​ <sup>1</sup> |  | ✅ |  |  |
@@ -90,13 +90,13 @@ AI service integration refers to internal connections to an Azure AI services mu
 
 ### Africa
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | South Africa North​ | ✅ | ✅ |  |  |
 
 ### Asia Pacific
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | Australia East​ ​| ✅ | ✅ | ✅ |  |
 | Australia Southeast​​​ |  |  | ✅ |  |
@@ -115,15 +115,15 @@ AI service integration refers to internal connections to an Azure AI services mu
 
 ## Azure Government regions
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | Arizona | ✅ |  | ✅ |  |
 | Texas |  |  |  |  |
 | Virginia | ✅ | ✅ | ✅ |  |
 
 ## Azure operated by 21Vianet
 
-| Region | AI service integration | Availability zones | Semantic ranker | Query rewrite |
+| Region | AI enrichment | Availability zones | Semantic ranker | Query rewrite |
 |--|--|--|--|--|
 | China East |  |  |  |  |
 | China East 2 <sup>1</sup> | ✅ |  |  |  |
