Tencent/DepthCrafter

Health v2 · maintenance only

Flags: prerelease_only no_license

Adoption not part of the score

1574 stars · 87 forks observed · 2026-08-28

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

DepthCrafter is a diffusion-based video depth estimation model from Tencent AI Lab that generates temporally consistent long depth sequences for open-world videos without requiring camera poses or optical flow. It ships as a Python library with inference scripts, a Hugging Face demo, and integrations into ComfyUI and Nuke.

Use cases

• estimate depth maps from monocular video
• generate temporally consistent depth sequences for visual effects
• get depth conditioning for video generation
• convert video depth to point clouds or 3D effects
• process long videos with segment-wise depth estimation
• export depth data as EXR for compositing pipelines

When to choose

• you need consistent video depth without camera pose or optical flow input
• you work with open-world videos of varying content and camera motion
• you want depth maps for VFX, compositing, or conditional video generation
• you need zero-shot depth estimation up to ~110 frames per pass

When to avoid

• you only need single-image depth estimation
• you need real-time depth on CPU or edge devices
• you require metric (absolute-scale) depth rather than relative depth
• you need a lightweight model without GPU or diffusion inference costs

Facets

library · maturity active

machine-learning deep-learning computer-vision video-processing image-processing computer-vision machine-learning artificial-intelligence python cross-platform depth-estimation video-depth diffusion-model monocular-depth cvpr-2025 research video gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Tencent/DepthCrafter")