The zipped version of docker file with details instructions available under this link:
VSCLIDockerfile.zip (please contact support to share this with you)

1. Overview

VStitcher-CLI requires:

Windows OS (Server 2022 recommended for containers)
VC++ Redistributables
StylezoneConnect (license login)
Optional GPU acceleration — only available on the host, not inside Windows containers
Mesa llvmpipe for CPU-only OpenGL rendering inside the container

Due to Windows container limitations:

👉 Windows containers do not get a real DirectX/WDDM GPU.
Unreal/DX12 rendering may fail inside the container.
CPU-only CLI exports (non-Unreal) can work using Mesa software rendering.

2. Installing NVIDIA “Gaming” Drivers on Cloud Providers

VStitcher (and most real-time 3D engines) require a WDDM GPU driver—not a headless TCC driver.

Cloud GPU instances often ship with the wrong driver mode. Below are exact methods to get the correct NVIDIA Gaming (WDDM) driver on AWS and GCP.

2.1 AWS EC2

Supported instance families

AWS instances supporting WDDM GPU drivers:

Instance family	GPU type	WDDM possible?
g4dn	Tesla T4	✔ Yes — via AWS Gaming Driver
g5	A10G	✔ Yes
g6	L4	✔ Yes
p2/p3/p4	V100/A100	✖ No (compute-only, TCC mode only)

How to download the AWS Gaming driver

AWS provides an S3 bucket with the WDDM gaming-certified driver.

Run this on your Windows Server EC2 instance (PowerShell):

Install-PackageProvider NuGet -Force 
Install-Module AWSPowerShell -Force 

$bucket = "nvidia-gaming" 
$prefix = "windows/latest" 
$local  = "$env:USERPROFILE\Desktop\NVIDIA" 

New-Item -ItemType Directory -Force -Path $local | Out-Null 

Get-S3Object -BucketName $bucket -KeyPrefix $prefix -Region us-east-1 |     ForEach-Object {
         if ($_.Size -gt 0) {
             $path = Join-Path $local $_.Key
             Copy-S3Object -BucketName $bucket -Key $_.Key -LocalFile $path -Region us-east-1
         }
     }

This downloads all required files, including the latest gaming WDDM driver.

Install it:

Start-Process "$local\GamingDriver.exe" -ArgumentList "/s" -Wait

Verify GPU is in WDDM mode:

nvidia-smi

Output must not say "TCC".
If it still does → reboot → reinstall driver.

2.2 Google Cloud Platform (GCP)

GCP provides GPU drivers through image bundles or manual installs.

Option A — Use prebuilt Windows image with GPUs

When creating a VM in GCP:

Choose GPU-supported Windows image under "Boot Disk" → "GPU-Optimized images"
Example:
Windows Server 2022 with NVIDIA drivers

These include WDDM gaming-capable drivers for supported GPUs (T4, L4, A10, etc.).

Option B — Manual installation

Download driver directly from NVIDIA Cloud Downloads:

$uri = "https://storage.googleapis.com/nvidia-drivers-us-public/GRID/GRID12.0/XXXX/NVIDIA-driver.exe" 
Invoke-WebRequest $uri -OutFile C:\NVIDIA-driver.exe 
Start-Process C:\NVIDIA-driver.exe -ArgumentList "-s" -Wait

(Replace path with the exact GPU type from GCP documentation.)

3. Architecture Limitations: No GPU Inside Windows Containers

Windows containers cannot access WDDM GPU acceleration
because DirectX requires a session 1 desktop environment, which containers do not have.

Even if you expose the GPU device:

--device class/5B45201D-F2F2-4F3B-85BB-30FF1F953599

You will still see inside the container:

Microsoft Basic Render Driver

This is expected.

Therefore:

Feature	Host Windows	Windows Container
DirectX 12	✔ Works	✖ Not available
Unreal rendering	✔ Works	✖ Fails
CPU rendering	✔ Works	✔ Works
Mesa OpenGL	N/A	✔ Works (software-only)

4. Preparing Build Context

Your build folder should contain:

Dockerfile 
installCli.ps1 
mesa/
   opengl32.dll
   libgallium_wgl.dll
   libglapi.dll 
CLI-TEST/
   cli_test.py
   WelcometoBW_Outfit.bw

5. Dockerfile (High-Level Summary)

Key operations:

Base image: mcr.microsoft.com/windows/servercore:ltsc2022
Install VC++ runtime
Download or copy VStitcher-CLI installer
Run silent setup
Copy Mesa DLLs
Add test scripts

6. Building the Container

From the folder with Dockerfile:

docker build --no-cache -t vstitcher-server:v1.0.0 .

7. Running the Container

Interactive PowerShell session

docker run --rm -it `
   --isolation process `
   --device class/5B45201D-F2F2-4F3B-85BB-30FF1F953599 `
   --entrypoint powershell.exe `
   vstitcher-server:v1.0.0

Inside container:

✔ All files are available
✖ GPU still appears as Microsoft Basic Render Driver (expected)

8. Testing Stylezone Login

Inside container PowerShell:

$stylezone = "$env:ProgramFiles\Browzwear\VStitcher-CLI\2025.2\StylezoneConnect.exe" 
& $stylezone --login --username="USER" --password="PASS" 

Write-Host "Exit code: $LASTEXITCODE"

Exit code 0 or blank = success.

9. Testing VStitcher-CLI Script

Example:

& "C:\Program Files\Browzwear\VStitcher-CLI\2025.2\VStitcher-CLI.exe" `
    --script C:\CLI-TEST\cli_test.py

Output is created under:

C:\CLI-TEST\output

Logs:

C:\Users\ContainerAdministrator\AppData\Local\Browzwear\log\VStitcher-CLI-54551\

10. Known Errors & Fixes

❌ DX12 not supported / Failing to choose D3D12 adapter

Cause: Windows containers cannot access WDDM GPU (DX12)
Fix:

Use host execution for DX12 workflows
Use Mesa (CPU) rendering inside container

❌ Stylezone license failure

Ensure outbound HTTPS allowed
Ensure correct credentials
Check DNS resolution

❌ Installer times out

Increase timeout in installCli.ps1

11. Recommendations

Goal	Recommended Approach
Full Unreal/DX12 rendering	Run VStitcher-CLI on host, not container
Scalable CPU-only CLI batch rendering	Use this Windows container
GPU acceleration inside container	Not supported by Windows OS today

12. Summary

This guide describes:

How to install correct NVIDIA Gaming drivers on AWS & GCP
How to build a VStitcher-CLI Windows container
How to run & test Stylezone authentication
How to execute CLI scripts
Limitations of DX12/GPU inside containers
Mesa CPU fallback setup

VStitcher-CLI — Local Installation & Usage Guide

How to Start Working With VStitcher Engine (VS-CLI)

VStitcher-CLI Windows Docker Image — Build & Run Guide