System requirements

This page lists what you need to run DocView Capture. Choose the column that matches how you plan to install it:

• Single workstation: one PC that both runs and uses DocView Capture.
• Server (IIS): Windows Server on your own network, used by several people from their browsers.
• Azure App Service: hosted in your own Azure subscription.

The parts of DocView Capture

DocView Capture has three programs. Each needs slightly different things.

All three share one SQL Server database. Page images and outputs are kept in one storage location: a folder or file share, or Azure Blob Storage. Images are never stored in the database.

Single workstation

A workstation installation serves only the local PC. For several users, install on a server with IIS instead. See Install on a single workstation.

Server installation with IIS

The web app and DocView Worker are self-contained for 64-bit Windows, so you don't need to install a separate .NET runtime. IIS does need the ASP.NET Core Module, which comes with the .NET 10 Hosting Bundle. See Install on IIS.

Important: HTTPS is required for scanning from other PCs. Plan a TLS certificate for the server's host name if operators will scan from their own desks.

SQL Server

• SQL Server Express is free and fine for getting started. It has a 10 GB database size limit. Because page images are stored outside the database, this goes a long way.
• Move to SQL Server Standard or Enterprise for large volumes, full recovery and backups.
• Both Windows Integrated Security and SQL authentication are supported.
• DocView Capture creates and upgrades its database schema itself when it starts, unless your DBA prefers to run the upgrade script.

Azure App Service

The Azure template creates everything you need:

• an App Service plan (Windows, P0v3 by default) for the web app and the DocView Worker WebJob;
• an Azure SQL database;
• a storage account for page images and outputs;
• a certificate that protects the shared key ring.

You need an Azure subscription and the Azure CLI. See Install on Azure.

Scanning PCs

Operators who scan need the DocView Scan Agent on the PC that the scanner is connected to.

• A Windows PC with a scanner that has a TWAIN or WIA driver.
• The agent installs for the operator's Windows account. It does not need administrator rights.
• It includes its own .NET runtime, so there are no other prerequisites. The download is about 175 MB; the installed agent about 90 MB.
• One download installs the 64-bit or the 32-bit agent, as the PC's scanner drivers need.

Operators download the agent from the Scan screen the first time they scan. See Scan Agent.

Operators who only import files, index or verify don't need the Scan Agent. They only need a web browser.

Supported file types

You can import:

• TIFF, single-page and multipage
• PDF, single-page and multipage
• JPEG
• PNG

Multipage PDFs and TIFFs are split into pages. By default, the largest single file accepted by browser upload and the REST API is 200 MB. You can change this with Maximum upload size (MB) under Administration › Settings.

OCR

• Tesseract runs on your own hardware and is the default. English and orientation data are included in the release.
• Azure AI Document Intelligence and Google Cloud Vision are optional. You supply your own key and pay the provider directly.

See OCR engines and languages.

Sizing tips

• OCR uses the most CPU. For higher volumes, add more DocView Workers rather than a bigger web server. Several DocView Workers share the same queue.
• In testing, one DocView Worker with 8 logical CPUs recognised about 15,700 pages per hour with Tesseract. Your results depend on page size, resolution and settings.
• Plan disk space for page images while batches are in progress, and for archived batches. Archived batches are deleted after 30 days by default.

Tip: Start small. A single server running the web app and one DocView Worker is enough for many organisations. See DocView Workers when you are ready to scale.

Sizing with DocView OmniRead

DocView OmniRead reads more accurately than Tesseract but needs more CPU. On a laptop without a graphics card, one installation read about 1,300 single-page colour scans per hour in accurate mode. That figure includes barcodes, tick boxes and the extraction rules. A server with 16 physical cores reads roughly 3,000 to 3,500 pages an hour. Check these figures with a pilot on your own pages and hardware.

Everything on one server. The web app, several DocView Workers and several OmniRead services can share one server:

• Memory. Allow about 2 GB for the web app, 1–2 GB for each DocView Worker and 2–4 GB for each OmniRead service. Add the memory you give SQL Server if it runs on the same server.
• OmniRead shares the server by itself. By default each OmniRead service sizes itself from the processor, keeps one core in eight free and reads at below-normal priority, so the server stays responsive. You can change the sizing and pool of each service in DocView (Settings › DocView OmniRead › Service settings). See DocView OmniRead.
• Instances. Several OmniRead instances can run on one server (DocView Setup asks for an instance name). They share the cores, so they add resilience and dedicated pools rather than speed.
• More servers for more speed or resilience. Add OmniRead on more servers and list every address under Administration › DocView OmniRead. DocView shares the work between them and skips a service that is not answering.
• DocView Workers. Run two or three DocView Workers so that OmniRead is kept busy. A DocView Worker handles one batch at a time.
• SQL Server. Use SQL Server Standard for more than a few thousand pages a day. The database of SQL Server Express is limited to 10 GB.
• Disk. A colour page at 200 dpi takes about 0.2 MB. With processed copies and exports, plan about 0.5 MB per page for as long as batches are kept.

See DocView OmniRead for its settings.