video generation (text-to-video / image-to-video)
An open-source diffusion model family (CogVideoX/CogVideo) for generating videos from text prompts or images, with training and inference code.
Same problem, same approach. Swapping one for another is a config change, not a rewrite.
Choose it for a larger model with stronger motion fidelity, accepting a custom Tencent license and heavier GPU requirements.
HunyuanVideo: A Systematic Framework For Large Video Generation Model
Pick it when you need fast, low-VRAM generation under a permissive Apache-2.0 license.
Official repository for LTX-Video
Swap in for a lighter, older diffusion T2V baseline with a smaller community and fewer controls.
VideoCrafter2: Overcoming Data Limitations for High-Quality Video Diffusion Models
Choose it for image-to-video with stronger subject preservation, but it needs the HunyuanVideo base weights.
HunyuanVideo-I2V: A Customizable Image-to-Video Model based on HunyuanVideo
Pick it for a permissively licensed (MIT) T2V/I2V family with multilingual text conditioning.
Kandinsky 5.0: A family of diffusion models for Video & Image generation
Choose it for a lighter T2V model that fits smaller GPUs, at the cost of Tencent's non-standard license.
HunyuanVideo-1.5: A leading lightweight video generation model
Pick it for a strong Apache-2.0 T2V/I2V family with a wide range of model sizes to trade off quality and VRAM.
Wan: Open and Advanced Large-Scale Video Generative Models
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Use it for real-time, long-horizon video where the seed's short clips are the limiting factor.
Helios: Real Real-Time Long Video Generation Model
Pick it when you need text-guided motion control while animating a single still image.
Official implementations for paper: LivePhoto: Real Image Animation with Text-guided Motion Control
Choose it when subject consistency from reference images matters more than broad text-to-video coverage.
Phantom: Subject-Consistent Video Generation via Cross-Modal Alignment
Pick it when you need synchronized audio and video generated by a single model.
A foundation model that generates synchronized video and audio in a single model
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
Unmaintained for roughly 1.7 years, so only take it if you specifically need its 2024 checkpoints.
Text and image to video generation: Kandinsky 4.0 (2024)
Unmaintained for two years, so only useful for legacy image-animation pipelines.
[ECCV 2024, Oral] DynamiCrafter: Animating Open-domain Images with Video Diffusion Priors
Does the same job, but ships as an app. Useful to a person, not swappable into a codebase.
A MidJourney-style web UI that calls Google's Imagen/Veo APIs rather than a self-hostable model.
Open-source clone of the MidJourney web interface featuring real AI image and video generation powered by Google's Gemini SDK. Use Imagen 4 to generate images and Veo 2 and 3 for image and text to video with audio.
A notebook/script demo chaining hosted video APIs instead of shipping a model.
Generate video from text using AI
These projects were analysed and named CogVideo among their alternatives. The relationship is not symmetric — how CogVideo rates them is a separate judgement, made when CogVideo is analysed in its own right.
calls CogVideo “Choose it when you need an alternative open-source text/image-to-video suite with Apache-2.0 and strong model maturity.”
Wan: Open and Advanced Large-Scale Video Generative Models
calls CogVideo “choose it for a mature, widely integrated text-to-video model that runs on lower-VRAM GPUs”
HunyuanVideo: A Systematic Framework For Large Video Generation Model
calls CogVideo “Choose when you need CogVideoX/CogVideo text/image-to-video model weights; Apache-2.0 but it does not cover Sana's image-generation side.”
SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer