# Preparation Before deploying Scenescape, prepare secrets, models, and container images for the target platform. Choose a source-build or prebuilt-image path below, then continue to [Deploy on Docker](./deploy-docker.md), [Deploy on Kubernetes](./deploy-kubernetes.md), or [Deploy the Demo](./deploy-demo.md). ## Prerequisites - Verify you meet the [System Requirements](../../get-started/system-requirements.md). - A checkout of the Scenescape repository (see [Get Scenescape](../../get-started/installation.md#step-1-get-scenescape)). ## Option A: Build from Source Run these commands from the repository root as a standard, non-root user: ```bash make ``` This is the same as `make build-core`: it generates secrets and certificates, builds the core microservice images (Manager, Scene Controller, Analytics, Auto Camera Calibration), and installs the default OpenVINO Model Zoo models. The build takes around 15 minutes, depending on the target machine. By default, the build runs in parallel with one job per processor. Adjust with `JOBS`, for example to build sequentially: ```bash make JOBS=1 ``` To also build the Tracker, Mapping, and Cluster Analytics images: ```bash make build-all ``` `make build-all` builds the Mapping image with the default MapAnything model only. To build both Mapping variants (MapAnything and VGGT), run: ```bash make -C mapping build-all ``` See [Build Mapping Service from Source](../../microservices/mapping-service/build-from-source.md) for the resulting image tags. ### Point Kubernetes Deployment To Prebuilt Images To build from source and make images available to an existing cluster, push them to a registry reachable from that cluster and override `repository` during Helm installation (see [Deploy on Kubernetes](./deploy-kubernetes.md)). For a local Kind cluster managed by this repository's tooling, see [Scenescape on Kubernetes](/kubernetes/README.md), which builds and loads images into the cluster automatically. ### (Optional) List Image Dependencies ```bash make list-dependencies ``` This generates dependency lists, with separate files for system packages and Python packages per microservice image. ## Option B: Use Prebuilt Images Prebuilt images are published on Docker Hub: - [Scenescape Manager](https://hub.docker.com/r/intel/scenescape-manager) - [Scenescape Controller](https://hub.docker.com/r/intel/scenescape-controller) - [Scenescape Analytics](https://hub.docker.com/r/intel/scenescape-analytics) - [Scenescape Autocalibration](https://hub.docker.com/r/intel/scenescape-autocalibration) - [Scenescape Tracker](https://hub.docker.com/r/intel/scenescape-tracker) - [Scenescape Cluster Analytics](https://hub.docker.com/r/intel/scenescape-cluster-analytics) - [Scenescape Mapping](https://hub.docker.com/r/intel/scenescape-mapping): `` and `-mapanything` contain MapAnything; `-vggt` contains VGGT ### Docker Prepare secrets and models; Docker Compose will pull the published images when you deploy: ```bash make init-secrets install-models ``` ### Kubernetes The chart pulls images from `repository` (default `docker.io`) using the published image names and the version in `kubernetes/scenescape-chart/Chart.yaml`, generated by `make -C kubernetes chart.yaml`. If your cluster can reach Docker Hub, no further image preparation is needed; continue to [Deploy on Kubernetes](./deploy-kubernetes.md).