1. Choose your model
Use your own ONNX model for detection and segmentation.
Your model.
Your target.
Your control.
ONNX aim assistance for Windows. Custom models, target continuity and configurable movement, in Free and Premium editions.
Public release 0.2.0Mission Control / Current release
2. Tune your workspace
Adjust tracking, compensation and output to your setup.
3. Stay in control
Clear controls, live preview and an Observe-only mode.
One foundation. Two editions.
Start with Free, or extend the same capture and tracking workspace with Premium.
Free edition
Capture, custom ONNX detection and instance segmentation, live visuals and ordinary tracking continuity. Includes controller and supported hardware output, profiles and the Official Aim Testing Suite. Selection stays fixed to Highest detection confidence.
Get Free for WindowsPremium edition
Everything in Free, plus alternate selection, bounded prediction, motion compensation, desktop overlay, the full training workflow, optional local Ultralytics training and API Suite control.
Explore PremiumBoth editions include twelve themes and a bundled guide in five languages.
Explore the tracking workspace
Native performance
C++ capture and ONNX inference for aim assistance, with DirectML default, CPU fallback, explicit CUDA and optional TensorRT RTX.
Your models
Load custom ONNX detection and instance-segmentation models with live annotated preview and mask visuals.
Target continuity
Tune capture zones, anchors, target retention, identity recovery and movement continuity through short detection gaps.
Connected control
Keyboard, mouse, controller and supported hardware movement output in both editions.
Practice before capture
The standalone Official Aim Testing Suite opens from Overview with saved preferences, custom characters and reports.
Build your setup with the community.
Get help, share feedback and discuss your setup with other Vapor Lock users.
Recent activity
Reported session and installation counts.
Loading activity
Waiting for the latest reported counts.
Only reported session and installation counts appear here. No license keys, identities, device details, images, model names or detection data are included.