Video Diffusion Models
29Generate, edit, and transform videos using cutting-edge diffusion architectures
Every major video diffusion architecture in C#. From Sora and CogVideoX for text-to-video, to AnimateDiff for image animation, to TokenFlow for video editing. Generate short clips, animate images, interpolate frames, and apply style transfer to video - all without Python.
Text-to-Video Generation
Generate videos from text descriptions with cinematic quality.
Sora
OpenAI spacetime transformer for up to 60-second realistic video generation.
CogVideoX
Expert transformer with 3D VAE for high-quality text-to-video.
HunyuanVideo
Tencent 13B parameter model for long-form video generation.
Kling
Kuaishou video generation with 3D spatiotemporal attention.
Veo
Google DeepMind high-fidelity video generation model.
Wan
Alibaba video generation with controllable motion dynamics.
LTX Video
Lightweight transformer for fast video generation.
Mochi
Genmo asymmetric DiT for high-quality motion generation.
Allegro
Rhymes AI text-to-video with temporal attention.
Open-Sora
Open-source Sora reproduction with DiT architecture.
Image-to-Video Animation
Animate static images into videos with controllable motion.
AnimateDiff
Motion module for animating any personalized text-to-image model.
AnimateDiff Lightning
Distilled AnimateDiff for 1-4 step animation.
Stable Video Diffusion
Stability AI image-to-video with multi-view generation support.
DynamiCrafter
Animate open-domain images with video diffusion priors.
I2VGen-XL
Alibaba high-quality image-to-video generation.
Video Editing & Transformation
Edit, restyle, and transform existing videos using text instructions.
TokenFlow
Consistent text-driven video editing via token propagation.
FateZero
Zero-shot video editing preserving temporal consistency.
Text2Video-Zero
Text-guided video generation/editing without training.
ControlVideo
Controllable video generation with spatial and temporal conditions.
VideoComposer
Compositional video generation with multiple conditions.
Video generation with AiModelBuilder
using AiDotNet;
// Train a video generation model with AiModelBuilder
var result = await new AiModelBuilder<float, float[], float>()
.ConfigureModel(new CogVideoX<float>(
numFrames: 49, fps: 8))
.ConfigureOptimizer(new AdamOptimizer<float>())
.ConfigurePreprocessing()
.BuildAsync(videoData, labels);
// Generate video from prompt embedding
var video = result.Predict(promptEmbedding);Start building with Video Diffusion Models
All 29 implementations are included free under Apache 2.0.