FOUNDATION / 03
SOFTWARE STACK
Build the software layer that connects your hardware to local models and creative tools. Set up the operating system, drivers, runtimes, containers, interfaces, and development utilities that make the studio reliable and easier to maintain.
START WITH THE OPERATING SYSTEM
Choose the platform that best supports the studio’s primary role.
WINDOWS
A practical choice for a main creative workstation using Blender, Unreal Engine, Adobe applications, and broadly supported desktop AI tools. Windows with WSL 2 can also provide access to many Linux-based development workflows.
LINUX
A strong choice for dedicated AI servers, containers, automation, and remotely managed services. Linux often provides direct access to AI development tools but may require more command-line setup and maintenance.
MACOS
A streamlined option for supported local models and creative applications on Apple silicon. Unified memory can be useful for certain workloads, but tool compatibility and upgrade options differ from systems built around dedicated GPUs.
INSTALL THE CORE LAYERS
Build a stable foundation before adding models and interfaces.
GPU DRIVERS
Install a current, compatible driver for the selected GPU and operating system. Confirm that the system recognizes the GPU correctly before installing AI applications.
MODEL RUNTIMES
Runtimes provide the engine used to load and run local models. Different tools may include their own runtime or connect to a shared service, so begin with only what your planned workflows require.
DEVELOPMENT TOOLS
Install Git for managing code and Python only when a tool or workflow requires it. Use isolated environments to prevent dependencies from one project from interfering with another.
CONTAINERS
Docker or another container platform can keep services isolated, reproducible, and easier to update. Containers are especially useful on a dedicated server, but they are not required for every desktop application.
CHOOSE HOW TO RUN YOUR TOOLS
DESKTOP APPLICATIONS
Installed applications provide the easiest starting point and usually manage much of their own setup. They work well for learning and interactive creative workflows.
WEB INTERFACES
Locally hosted browser interfaces can provide more control, extensions, and reusable workflows. They may run directly on the workstation or connect to a service on another machine.
LOCAL SERVICES
Persistent model servers make AI capabilities available to multiple applications, automations, and devices through an API. This is useful when building an integrated studio rather than using each tool separately.
COMMAND LINE
Command-line tools offer direct control, scripting, and automation. They are powerful for advanced workflows but are not necessary for every user or every task.
KEEP THE STACK MANAGEABLE
A smaller, documented stack is easier to trust and maintain.
ASSIGN EACH TOOL A ROLE
Choose a primary tool for each function before installing alternatives. Add another only when it provides a capability your current stack lacks.
SEPARATE ENVIRONMENTS
Keep project dependencies isolated with virtual environments, containers, or application-managed installations. Avoid placing every Python package into one shared system environment.
ORGANIZE FILES
Create predictable locations for applications, models, caches, workflows, outputs, and long-term storage. Consistent paths make backups, migrations, and troubleshooting easier.
CONTROL UPDATES
Record working versions and avoid updating every component immediately. Test important updates before changing a stable production workflow.
PROTECT THE LOCAL STUDIO
NETWORK ACCESS
Keep services restricted to the local machine unless another trusted device needs access. Do not expose local AI interfaces directly to the public internet.
CREDENTIALS
Store API keys, passwords, and access tokens outside project files and shared repositories. Use environment variables or a credential manager when supported.
BACKUPS
Back up configurations, workflows, custom models, scripts, and project files. Downloaded base models can often be replaced; original work and configuration history cannot.
RECOVERY
Document installation steps and working versions so the system can be rebuilt after a failed update, drive replacement, or operating-system change.
CREATE YOUR SOFTWARE PLAN
Define the stack before beginning a large installation session.
Record the following:
PLATFORM
Choose the operating system for each workstation or server and note whether WSL, virtual machines, or containers will be used.
CORE SERVICES
List the model runtimes, local servers, interfaces, and creative applications required by the system brief.
FILE STRUCTURE
Define where models, caches, workflows, outputs, projects, and backups will live.
ACCESS
Decide which tools remain available only on the host machine and which trusted devices may connect over the local network.
MAINTENANCE
Record version information, update procedures, backup locations, and the steps needed to restore the stack.
BUILD IN STAGES
Install and test one layer at a time:
01 — SYSTEM
Update the operating system, install GPU drivers, and confirm that the hardware is recognized.
02 — RUNTIME
Install one model runtime and verify that it can use the intended GPU or accelerator.
03 — FIRST MODEL
Run one small, well-supported model before adding more tools or larger downloads.
04 — INTERFACE
Add the desktop or web interface that will control the first working model.
05 — CREATIVE TOOLS
Connect Blender, Unreal Engine, image, video, or other production applications only after the core stack is stable.
06 — AUTOMATION
Add APIs, MCP connections, and persistent services after the individual components work reliably.
NEXT STEP
Choose the models that will power the studio.
Match model capabilities and sizes to your hardware, workflows, quality requirements, and licensing needs before building a large local library.