Table of Contents

Property MaximumImageDetail

Namespace
LMKit.Document.Parsing
Assembly
LM-Kit.NET.dll

MaximumImageDetail

The most pixels a reading model is given for a page, a figure or a second reading. null (the default) decides from the hardware.

public ImageDetail? MaximumImageDetail { get; set; }

Property Value

ImageDetail?

Remarks

The effort level decides the resolution each page and figure is read at, up to Maximal for the pages that need a second look. On a graphics device that costs seconds per page. On a processor without one it costs minutes: the vision encoder's work grows faster than the number of pixels, because every patch of the image attends to every other.

With the default null, a parser whose page model runs its vision encoder on the CPU reads at most at Low, several times faster per page, at the price of the finest print: small text, dense tables and tight chart labels lose detail. On a graphics device nothing is lowered.

A stated value applies on every device, lowering every reading above it: a speed setting for a large batch on a graphics device, or a way back to the full resolution on the CPU with Maximal, which lowers nothing. A second reading the ceiling would serve at the budget of the first is not made. The setting is read when a parse starts.

Share