Unreal Profiling With Python

Jupyter Notebook has many useful tools to visualize data gathered from Unreal Engine games

This repository has code to start gathering and visualizing data from UE games via Session Fronted and Jupyter Notebook

How to start using

run get_dependencies.bat to setup venv with dependencies based on python that packaged with unreal engine installation

Running profiling_project

Profiling_project is a container Project for ProfilingBPLibrary cpp plugin that exposes https://docs.unrealengine.com/5.1/en-US/API/Runtime/SessionServices/ISessionInstanceInfo/ExecuteCommand/ to Blueprints/Python

• run profiling_project/profiling_project.uproject

• Enable Remote Execution in Unreal

  Project Settings -> Plugins -> Python -> Python Remote Execution -> Enable Remote Execution = True

Jupyter Notebook Setup

Minimal code for unreal engine connection and command sending:

from remote_execution import RemoteExecution
remote_exec = RemoteExecution()
remote_exec.start()
for node in remote_exec.remote_nodes:
    remote_exec.open_command_connection(node.get("node_id"))
remote_exec.run_command(f"unreal.ProfilingBPLibrary.send_command('stat unitgraph')")

example in test_example.ipynb

How to run game so it's visible in Session Frontend

Add argument -Messaging when you run builds

example command to run on windows - profiling_project\Script\run_on_external_machine.bat

Machines on local network should be added in

Project Settings -> Plugins -> UDP Messaging -> Static Endpoints

or Engine.ini

[/Script/UdpMessaging.UdpMessagingSettings]
StaticEndpoints=192.168.**.***:6666

Use -execcmds="session AUTH username_of_controlling_pc" to unlock machines in local network for commands

utils - session_ctrl

Designed to be used with Jupyter Notebook to automate gathering and visualization of Unreal Engine profiling commands like profilegpu, StartFpsChart and StopFpsChart

usage example in notebook test_example.ipynb