SkyReelsV1Model<T>

Models & Types · AiDotNet.Diffusion.Video

SkyReels V1 human-centric video generation model.

Inheritance: DiffusionModelBase<T> → LatentDiffusionModelBase<T> → VideoDiffusionModelBase<T> → SkyReelsV1Model<T>

For Beginners

SkyReels V1 is a video generation model trained on 10 million film clips, specializing in realistic human-centric content. It excels at generating natural human motion, facial expressions, and interactions in cinematic style.

How It Works

References:

  • Paper: “SkyReels V1: Human-Centric Video Foundation Model” (Kunlun Tech, 2025)

SkyReels V1 is a specialized human-centric video model built by fine-tuning HunyuanVideo on approximately 10 million high-quality film and television clips. It focuses on realistic human portrayals with cinematic quality, particularly excelling at facial expressions, body movements, and human-environment interaction.

Technical specifications:

  • Architecture: DiT (HunyuanVideo-based) + Human-Centric Fine-tuning
  • Latent channels: 16
  • Default: 49 frames at 24 FPS
  • Supports I2V: Yes | T2V: Yes | V2V: No

Example

var options = new DiffusionModelOptions<float> { LatentChannels = 16, DefaultInferenceSteps = 30 };
var noise = Tensor<float>.CreateRandom(new[] { 1, 16, 49, 90, 160 });
var trainX = Tensor<float>.CreateRandom(4, 8);
var trainY = Tensor<float>.CreateRandom(4, 2);
var result = new AiModelBuilder<float, Tensor<float>, Tensor<float>>()
    .ConfigureModel(new SkyReelsV1Model<float>(options: options))
    .Build(trainX, trainY);
var video = result.Predict(noise);

Constructors

Constructor Summary
SkyReelsV1Model(NeuralNetworkArchitecture<T> architecture, DiffusionModelOptions<T> options, INoiseScheduler<T> scheduler, DiTNoisePredictor<T> predictor, TemporalVAE<T> temporalVAE, IConditioningModule<T> conditioner, int defaultNumFrames, int defaultFPS, int? seed) Initializes a new instance of SkyReelsV1Model with full customization support.

Properties

Property Summary
IConditioningModule<T> Conditioner { get; }
INoisePredictor<T> NoisePredictor { get; }
IVAEModel<T> TemporalVAE { get; }
IVAEModel<T> VAE { get; }
bool SupportsImageToVideo { get; }
bool SupportsTextToVideo { get; }
bool SupportsVideoToVideo { get; }
int LatentChannels { get; }

Methods

Method Summary
ModelMetadata<T> GetModelMetadata()