The following are the three common workflows for cloud-based image management and publication:
- Generate tile cache and serve
- Generate tiled imagery and serve
- Generate dynamic imagery and serve
In all three workflows, mosaic datasets are used as the primary means to manage imagery. Mosaic datasets are created in a geodatabase that reference the imagery and define functions to be applied on the imagery to create a virtual mosaic. Users can access mosaic datasets directly in ArcGIS Pro, or the imagery can be exported to suitable formats; such is the case when serving tile cache, tiled imagery, or raster datasets. When creating a mosaic dataset for large imagery collections that will be served as a single layer, the mosaic dataset is served directly.
Tile cache
Tile cache stores images as preprocessed tiles in a defined map projection and at defined scales. It is typically used as basemap imagery in applications. You can also publish an elevation surface as a tile cache to use in 3D scenes. Tile cache can be stored as a TPK file, a TPKX file, or as ready-to-serve content. The TPK/TPKX-or tile package-file format is supported across ArcGIS and is typically used for basemaps.
Advantages
This option is simple, inexpensive, and doesn't require having a server or storing your data in the cloud.
Generate tile cache and serve
Generating a tile cache is simple to do even on large datasets. There are a number of caching tools in ArcGIS Pro to generate, manage, export, and share a tile cache, and serve it to ArcGIS Online, ArcGIS Enterprise, or ArcGIS Image Dedicated. A tile cache can also be brought back into ArcGIS Online as a raster, as can a TPKX, which is a packaged version of that tile cache. The custom Raster Tile Cache tools are another option—this includes four simple tools to create, package, publish, and update tile cache.
While a tile cache doesn't include any metadata about the imagery, you can create a feature service from the mosaic dataset metadata, which you can associate with your tile cache service to make it possible to access basic information. Esri's World Imagery basemap utilizes this strategy. More information is available in the Generate and Publish Tile Cache workflow.
You can also use ArcGIS Image Dedicated to generate a tile cache. You will need to use ArcGIS Pro to create a mosaic dataset and then use the Convert and Persist toolbox to scale your data into an optimized format and write it to a cloud storage location. The Convert and Persist Python toolbox is a part of the Processing and Analysis subscription for ArcGIS Image Dedicated.
Image services
Image services provide users a way to stream tiled or dynamic imagery layers as services in the cloud either hosted by Esri using ArcGIS Image for ArcGIS Online, in your own AWS or Azure cloud using Image Dedicated, or in your organization's enterprise cloud-based storage solution using ArcGIS Image Server. These image services allow for more than just visualization; with them, you can perform and scale up image and raster analytics using different deployment options.
Tiled imagery
Tiled imagery layers are image services that you can generate from one image or multiple images. If you create a tiled imagery layer with multiple images, these images mosaic together as one image. With tiled imagery, you can serve imagery with any number of bands, bit depths, and in a projection. Tiled imagery is primarily stored as a Cloud Raster Format (CRF) and allows you to access source pixel values and perform analyses using raster analysis tools. And unlike tile cache, you are able to continuously make visualization changes to tiled imagery.
Advantages
This option performs well, preserves metadata, maintains full pixel fidelity, and is analysis ready.
Generate tiled imagery layers and serve
Generating and serving tiled imagery is very similar to tile cache in that it can serve static tile data and can be shared publicly on ArcGIS Online. However, it can also support imagery with more than three bands and multidimensional data, which makes it ideal for visualization and image analysis. When imagery files are typically hosted with ArcGIS Image for ArcGIS Online, they are stored, or converted, to CRF files. With ArcGIS Image for ArcGIS Online, you don't need your own infrastructure. You are able to directly upload your imagery or raster data from the web viewer or author from ArcGIS Pro by creating mosaic datasets that you can persist as tiled imagery. From there, you can serve your imagery as a hosted tiled imagery layer. Since the format is optimized, the web client can also perform client-side rendering and re-projection on the layer.
You can publish tiled imagery datasets in CRF format as tiled imagery services with ArcGIS Image Dedicated. When using Image Dedicated, the imagery data source should reside in cloud storage—either Microsoft Azure or Amazon Web Services (AWS). Typically, an ArcGIS Cloud Storage (ACS) connection file is used to access the imagery data in cloud storage and create the tiled imagery service.
Using ArcGIS Image Server, you can author and publish tiled imagery layers using ArcGIS Pro or ArcGIS Enterprise and then publish the managed imagery as image services by reference or by value.
Note:
You will need an ArcGIS Image for ArcGIS Online license to publish to ArcGIS Online as hosted tiled imagery layers.
Dynamic imagery
Dynamic imagery is a collection of images that are dynamically mosaicked together on-the-fly by the server. Dynamic imagery layers are useful for overlapping or multitemporal imagery. Multiband and multidimensional data are also supported; with multiband imagery, you are able to access pixel values and band information with ArcGIS Image for ArcGIS Online.
Advantages
This option lets you use processing templates, control input image visualization, and supports on-the-fly analysis.
Generate dynamic imagery layers and serve
In ArcGIS Online, you will find that you are unable to host and stream this type of layer with the public. However, you can serve this imagery using Image Dedicated or ArcGIS Image Server as a publicly accessible image service. With ArcGIS Image Dedicated, you can publish dynamic imagery as dynamic image services with mosaic datasets built from rasters. When using Image Dedicated, the imagery data source should reside in cloud storage—either Azure or AWS.
You can also host your dynamic imagery using ArcGIS Enterprise with ArcGIS Image Server; this enables a user to upload imagery to ArcGIS Enterprise and have it served back as a dynamic image service accessible from the web.
This solution can be implemented on-premises or in the cloud. The source imagery can be stored locally or it can be referenced imagery in the cloud. ArcGIS Enterprise with ArcGIS Image Server can be implemented either on-premises or in the cloud. For organizations that don't intend to use all the capabilities of ArcGIS Enterprise, including a portal, it is possible to install and use ArcGIS Pro 2.4 and later and a stand-alone ArcGIS Image Server in the cloud to publish dynamic imagery. In this workflow, you'll store imagery in the cloud, serve it as a dynamic image service using ArcGIS Pro and ArcGIS Image Server, and then either access the service directly using the REST URL, or register the service with ArcGIS Online and then use ArcGIS Online as the portal for that image service (both the Landsat and Sentinel-2 image services from Esri are managed this way).
Similar to tiled imagery layers, you can directly upload your imagery or raster data from the web viewer and host. However, keep in mind that storage costs for dynamic imagery depend on the size and number of images. Review help documentation on understanding credits to better understand imagery storage options and credit costs with ArcGIS Image for ArcGIS Online.
Note:
You can also use the Publish Image Service tool included with MDCS to publish image services to a stand-alone ArcGIS Image Server using ArcGIS Pro 2.3 and later. In addition to publishing image services, it also creates the imagery layers for you in ArcGIS Online, and allows you to define whether you want to use dedicated or shared instances.