Enum ImageDetail

Namespace

LMKit.Inference.Vision

Assembly

LM-Kit.NET.dll

Specifies the level of detail used when processing images for vision models. Controls the maximum pixel budget allocated to an image, which directly affects token consumption and visual fidelity.

public enum ImageDetail

Fields

Minimal = 0

Lowest resolution. Constrains the image to the model's minimum pixel budget, resulting in the fewest vision tokens per image.

Low = 1

Reduced resolution that limits the pixel budget below the default, trading visual fidelity for lower token usage.

Standard = 2

Balanced resolution that provides a moderate pixel budget, offering a good compromise between fidelity and token cost.

High = 3

High resolution. Uses a large pixel budget for detailed image analysis. This is the default level used when no explicit detail is specified.

Maximal = 4

Highest resolution. Allocates the maximum pixel budget supported by the model, maximizing visual fidelity at the cost of higher token consumption.