IVideoSuperResolution<T>

Interfaces ยท AiDotNet.Interfaces

The video super-resolution contract: upscale a low-resolution frame sequence.

Known implementations: DiffusionVideoSuperResolutionBase<T>, VideoSuperResolutionBase<T>

For Beginners

video super-resolution takes a low-resolution clip and produces a higher-resolution one, using neighbouring frames for detail a single image cannot supply.

How It Works

This interface exists because the capability is NOT tied to one base class. Most models in the library inherit VideoSuperResolutionBase, but diffusion-based ones (Stream-DiffVSR, StableVideoSR, MGLD-VSR, DOVE, SeedVR, FlashVSR) genuinely need the latent/denoising machinery on VideoDiffusionModelBase instead โ€” and C# permits one base class. Declaring the capability as an interface lets both hierarchies be recognised as video super-resolution by consumers and by the test-scaffold generator, rather than the generator matching on a base-class NAME prefix, which silently excludes any model whose base is named differently.

Properties

Property Summary
int ScaleFactor { get; } Gets the spatial upscaling factor (e.g.

Methods

Method Summary
Tensor<T> EstimateFlow(Tensor<T> frame1, Tensor<T> frame2) Estimates the optical-flow field between two frames.
Tensor<T> Upscale(Tensor<T> lowResFrames) Upscales a low-resolution frame sequence.