Mira-Scene

Mira-Scene reconstructs an editable 3D scene from a single image. The full pipeline combines instance segmentation, monocular depth estimation, canonical coordinate map (CCM) and occupancy prediction, per-object mesh generation, support-aware scene assembly, and environment-map generation.

This repository contains the released Mira-Scene CCM diffusion checkpoint. The project code and complete inference instructions are available in the Mira-Scene GitHub repository.

Model details

The checkpoint predicts a pixel-aligned canonical coordinate map and a sparse voxel representation for each segmented object. It is packaged as a CCMVoxelPipeline and contains the following components under pipeline/:

• a DINOv2-with-registers image encoder;
• a flow-matching scheduler;
• a CCM/voxel diffusion transformer;
• a sparse-structure VAE; and
• the image preprocessor configuration.

The checkpoint is one stage of Mira-Scene, not a standalone image-to-GLB model. A scene image and per-instance masks are prepared by the segmentation stage; a supported mesh backend and the scene-construction stages consume its outputs.

Download

Install and authenticate the Hugging Face client if necessary, then download the checkpoint:

python -m pip install -U huggingface_hub
hf auth login
hf download Yang-Tian/Mira-Scene \
  --include "pipeline/*" \
  --local-dir checkpoints/mira-scene

Set checkpoints.ccm in infer_scripts/config/local.yaml to the downloaded pipeline directory:

checkpoints:
  ccm: /absolute/path/to/checkpoints/mira-scene/pipeline

The full pipeline also uses third-party segmentation, depth, and mesh checkpoints. See the environment and checkpoint guide for the exact model IDs, download commands, and configuration keys.

Inference

After installing the stage-specific environments and configuring all selected backends, run:

python infer_scripts/pipeline.py \
  --input /path/to/image_or_directory \
  --output /path/to/results \
  --config infer_scripts/config/local.yaml

To run only the CCM stage on already prepared cases:

python infer_scripts/2_inference_CCM.py \
  --demo_dir /path/to/prepared_cases \
  --output_dir /path/to/results \
  --ckpt_dir /absolute/path/to/checkpoints/mira-scene/pipeline \
  --use_cropped_condition

Refer to the project inference guide for the required input layout and the complete stage-by-stage workflow.