WorldFoundry
English中文

Prior Depth Anything

Method overview: coarse metric alignment of any depth prior, then implicit RGB-conditioned refinement through a frozen monocular depth estimator.

Prior Depth Anything: depth-completion and metric-depth-estimation; Verified.

Edit this pageGitHubMarkdown

Prior Depth Anything is a depth-completion prior: it takes any coarse or sparse metric-depth prior plus an image and refines them into a dense, sharper metric depth map. In this catalog it is verified — the Workspace loaded the official checkpoint and refined a real Depth Anything 3 metric-depth condition on 2026-07-11.

Official paper teaser: Prior Depth Anything fuses dense geometric estimates with sparse metric measurements.
Official paper teaser: Prior Depth Anything fuses dense geometric estimates with sparse metric measurements.
Method overview: coarse metric alignment of any depth prior, then implicit RGB-conditioned refinement through a frozen monocular depth estimator.
Method overview: coarse metric alignment of any depth prior, then implicit RGB-conditioned refinement through a frozen monocular depth estimator.

The recorded paper is arXiv 2505.10565.

Project page: prior-depth-anything.github.io. Code: SpatialVision/Prior-Depth-Anything.

Build your run

Pipeline
PriorDepthAnythingPipelineworldfoundry.pipelines.geometry_priors.pipeline_geometry_priors
TaskDefault default
Environmentworldfoundry-unified-cu128
DeviceCUDA 12.8
Runnerworldfoundry.pipeline
Loadingfrom_pretrained
Invocationnative_or_standard_invocation
worldfoundry-eval run \
  prior-depth-anything \
  --pipeline.task-profile default \
  --json

What it is for

Use prior-depth-anything for catalog geometry inference. Choose it when the artifact is a reconstruction, not a generated video.

The catalog records a prompt-conditioned depth-completion design: the model consumes prompt metric depth alongside the image (checkpoint Rain729/Prior-Depth-Anything, apache-2.0). It is registered at base_models/three_dimensions/depth/priorda and shared by higher-level geometry pipelines without being nested under them.

Catalog tasks include depth-completion, metric-depth-estimation, and geometry-prior. Catalog aliases include priorda and prior-depth-anything-prior. Documented modalities: image in, depth out.

Runtime contract

Optional input_path, prompt_depth_path, prior_depth_path, execute, and max_points; artifacts prior-depth-anything.npz.

CLI task profiles: default. WorldFoundry binds PriorDepthAnythingPipeline.

Family status: Verified.

It is integrated as an independent geometry prior under worldfoundry/base_models and exposed through the shared geometry-prior pipeline/operator contract. Higher-level pipelines (e.g. ViPE) import it from this canonical owner instead of keeping private copies. Pair it with a metric source (e.g. depth-anything-v3-prior) whose output it refines.

Integrated as an independent depth-completion prior at worldfoundry/base_models/three_dimensions/depth/priorda.

Running in WorldFoundry

The catalog launch is worldfoundry-eval run prior-depth-anything --pipeline.task-profile default. The command builder above switches --pipeline.task-profile among recorded tasks and still emits worldfoundry-eval run.

It runs in the unified environment worldfoundry-unified-cu128 (Python 3.11, CUDA 12.8).

Recorded weight repositories include Rain729/Prior-Depth-Anything. Prepare the environment and check local assets with bash scripts/inference/prepare_model_infer.sh prior-depth-anything. Add --download to fetch missing public checkpoints.

Recorded licenses include apache-2.0; status: Verified.