Skip to content

Intel RealSense Camera Operator#

Authors: Holoscan Team (NVIDIA)
Supported platforms: x86_64, aarch64
Language: C++
Last modified: August 5, 2025
Latest version: 0.1.0
Minimum Holoscan SDK version: 2.1.0
Tested Holoscan SDK versions: 2.1.0
Contribution metric: Level 1 - Highly Reliable

Overview#

The RealSense Camera Operator captures synchronized color and depth frames from Intel RealSense cameras using the RealSense SDK. This operator provides real-time streaming capabilities with configurable resolution and frame rates, making it ideal for computer vision applications, robotics, and 3D reconstruction tasks.

Features#

  • Dual Stream Capture: Simultaneously captures color (RGBA) and depth (Z16) video streams
  • Synchronized Frames: Ensures temporal alignment between color and depth data
  • GPU Memory Management: Efficient CUDA memory allocation and transfer
  • Camera Intrinsics: Provides camera calibration parameters for both streams
  • Configurable Resolution: Supports various resolution and frame rate combinations
  • Real-time Processing: Optimized for low-latency streaming applications

Usage#

Basic Example#

#include "holoscan/holoscan.hpp"
#include "realsense_camera.hpp"

class MyApp : public holoscan::Application {
 public:
  void compose() override {
    auto realsense = make_operator<holoscan::ops::RealsenseCameraOp>(
        "realsense",
        Arg("allocator", make_resource<holoscan::UnboundedAllocator>("pool")));

    add_operator(realsense);
  }
};

Advanced Configuration#

auto realsense = make_operator<holoscan::ops::RealsenseCameraOp>(
    "realsense",
    Arg("allocator", make_resource<holoscan::UnboundedAllocator>("pool")));

Please refer to the Intel RealSense Camera Visualizer reference application for an example usage.

Outputs#

The operator provides the following outputs:

Video Buffers#

  • color_buffer: RGBA8 color video stream (1280x720 @ 30fps by default)
  • depth_buffer: Z16 depth video stream (1280x720 @ 30fps by default)

Camera Models#

  • color_camera_model: Intrinsic parameters for the color camera
  • depth_camera_model: Intrinsic parameters for the depth camera

Camera Model Structure#

Each camera model contains:

  • Dimensions: Width and height in pixels
  • Focal Length: fx, fy in pixels
  • Principal Point: ppx, ppy in pixels
  • Distortion Type: Currently set to Perspective

Configuration#

Default Settings#

  • Color Stream: 1280x720, RGBA8 format, 30fps
  • Depth Stream: 1280x720, Z16 format, 30fps
  • Alignment: Depth frames aligned to color stream

Supported Formats#

  • Color: RGBA8, RGB8, BGR8, YUYV
  • Depth: Z16, Y16, Y8