Follow the image and controller paths
The image path begins at a source the computer can capture, such as a capture device or a supported window. Gtuner IV can display that source and run a compatible computer vision script. The controller path is separate. A connected Titan Two exposes its inputs and sends the configured output to the gaming system.
The architecture lets someone build a read-only experiment, for example counting the appearances of a menu indicator in recorded footage. Verify that experiment on a saved clip before relying on live capture. Footage recorded at one resolution or with a different interface scale can change which pixels a model sees.
Questions to ask before buying a toolkit
Ask for the host operating system, exact software versions, expected video source, supported game builds and the contents of the plan. A controller adapter does not include the computing resources or files a separate model needs. The hardware compatibility list also cannot establish a model’s accuracy.
When evaluating a demo, look for missed detections as well as successful ones. Cropped highlights hide false positives, capture delay and behavior when a menu appears. Check the game’s rules before using anything that changes input during play. A test built around a local clip gives you a way to inspect the vision side without sending automated actions into a game.
Check what the video source exposes
Use a short local test clip that shows the visual element you want to study. Check that the frame has the intended resolution, aspect ratio and visible region. Before sharing a clip for troubleshooting, crop out account names and unrelated windows. Describe the expected detections in plain language so another person can evaluate the same footage.
Sources
Titan Two AI publishes these guides independently. ConsoleTuner makes the Titan Two device and Gtuner IV.