Sycrad logo

Frequently Asked Questions

Answers to common questions about installing, running, and evaluating Sycrad.

01

How do I run Sycrad on Windows?

During pre-alpha, Sycrad Windows builds are not yet code-signed. If Microsoft Defender SmartScreen intercepts the installer or executable, click More info and select Run anyway. This override is only required on first launch. Sycrad requires 64-bit Windows 11 with DirectX 12 support.

02

How do I run Sycrad on macOS?

During pre-alpha, Sycrad macOS builds are not yet code-signed. If Gatekeeper intercepts the installer or application bundle, open System Settings > Privacy & Security and select Open Anyway. This override is only required on first launch. Sycrad requires macOS 15 or later on Apple silicon.

03

How do I run Sycrad on Linux?

During pre-alpha, Sycrad Linux builds are not yet available. Development and maintenance are currently focused on macOS and Windows. Native Linux support is on our roadmap for subsequent releases.

04

Which datasets and file formats are supported?

Sycrad supports KITTI Raw, nuScenes v1.0, and Waymo Open Dataset (v2 Parquet) for complete annotation, interpolation, and SLAM workflows. PCD and LAS/LAZ files are supported strictly for visualization. Future updates will introduce annotation workflows for PCD and LAS/LAZ alongside broader dataset coverage, including autonomous vehicle formats like Argoverse and robotics perception datasets on our roadmap.

05

What are the minimum hardware requirements?

Sycrad requires DirectX 12 on Windows and an Apple M-series chip on macOS. System RAM and VRAM requirements depend directly on the dataset loaded into the viewport. Rendering performance scales with the GPU, while data processing and computation rely on multi-threaded CPU performance.

06

Does my data leave the machine?

No. Your data never leaves your local machine; all rendering and processing run entirely on local hardware. Non-enterprise plans require an internet connection solely for license validation. In University and Team plans, only cuboid annotations sync through the cloud to coordinate work between users, while Enterprise deployments run completely offline in air-gapped environments.

07

How does Sycrad compare to cloud-based annotation platforms?

Cloud-based platforms require uploading gigabytes of point cloud data over the internet, incurring recurring bandwidth costs, egress fees, and data privacy risks. Browser-based interfaces also face memory and rendering constraints that limit sequence playback and responsiveness. Sycrad runs as a native desktop application directly on your local workstation. Point clouds remain entirely on local storage, eliminating data transfer delays, cloud egress bills, and browser performance bottlenecks.

08

Can Sycrad run in completely air-gapped or isolated environments?

Yes. The Enterprise edition is designed specifically for air-gapped, isolated, and strictly regulated enterprise environments. It operates completely offline with zero outbound network communication or external telemetry. All point clouds, annotations, and project files remain entirely within your private infrastructure.

09

What is the Interpolation workflow and how does it work?

The Interpolation workflow accelerates sequence labeling by automatically filling intermediate frames between user-defined keyframes. Annotators set keyframes by placing bounding cuboids at the start and end of a frame range for one or multiple objects. The interpolation algorithm then tracks each object across the sequence and populates every frame in between, preserving box dimensions and adjusting to movement. This turns repetitive frame-by-frame drawing into rapid, multi-object track generation in seconds.

10

What is the SLAM workflow and how does it work?

The SLAM workflow merges consecutive frames into a single dense scene. Accumulating points across frames makes objects significantly clearer and faster to annotate than inspecting individual frames. For dynamic objects, the pipeline detects moving points, clusters them into candidate objects, and tracks them into continuous trajectories, so a single cuboid follows the object across the entire sequence. For static objects, Sycrad automatically detects only the frames where they remain visible and adds them accordingly. This turns sequence labeling into a fast, single-action workflow.