alibaba/Tora

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

1241 stars · 60 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Tora is Alibaba's official implementation of a trajectory-oriented Diffusion Transformer (DiT) for controllable video generation, integrating textual, visual, and trajectory conditions. It lets users generate videos whose object motion follows user-specified trajectories, with support for text-to-video and image-to-video via SAT and diffusers weights.

Use cases

• generate videos following a drawn object trajectory
• text-to-video generation with motion control
• image-to-video generation with controllable dynamics
• research on trajectory-conditioned diffusion transformers
• create videos where objects move along custom paths

When to choose

• you need fine-grained control over object motion paths in generated video
• you want a research-grade DiT video model with published weights
• you're exploring trajectory-conditioned video generation

When to avoid

• you need a polished end-user video creation app
• you lack a GPU or don't want to run large diffusion models locally
• you need simple text-to-video without motion control

Facets

library · maturity active

video-processing machine-learning deep-learning image-processing deep-learning computer-vision artificial-intelligence python diffusion-transformer text-to-video image-to-video trajectory-control motion-control video-generation research-code cvpr-2025 video gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="alibaba/Tora")