Intel® Perceptual Computing SDK

SDK R8 installer trouble

Hello all,

I've got the SDK R8 offline installer problem, so it is quite helpful if somebody gives me an advice.

I installed perceptual SDK R8, but found my program which runs in R7 crashed, and even sample program crashed too. So I uninstalled R8 and tried to re-install R7, but somehow the installer thinks R8 is still on my PC and did nothing because it thinks newer version is already installed. I guess R8 installer does not clean it's information on my PC, so I'd like to know how I can clean up R8 information on my PC.

Usage of MapColorCoordinatesToDepth

Hello support team,

I really appreciate if you show an example of MapColorCoordinatesToDepth usage because it's arguments are different from SDK reference manual, so I don't know what I should pass to this function. This function seems to require below arguments in C# case, but I have no idea what these parameters are.

1. PXCMImage image <= Is this a color image or depth image?

2. PXCMPointF32[] uvInv <= What is this?

3. uint uvInvStep <= What is this?

4. PXCMSizeU32 uvInvRoi <= What is this?

work with opencv

   I am a green hand to the RealSense. But I really interested in it's great abilities to interact with human. 

  Now I want to get the depth image by Senz3d and deal with these images by OpenCV. But how can I post these data gotten by Senz3d to OpenCV? Are there any functions in SDK to finish my tasks?

  It is my pressure to get your answers.

Capture Image from Depth/Color Stream

Hi all, I'm wondering how do I capture images from a depth or color stream? I'm using the sample C++ code from PCSDK

#include <stdio.h>
#include <conio.h>
#include <windows.h>
#include <wchar.h>
#include <vector>
#include <tchar.h>

#include "pxcsession.h"
#include "pxcsmartptr.h"
#include "pxccapture.h"
#include "util_render.h"
#include "util_capture_file.h"
#include "util_cmdline.h"

Accuracy improvement request for PoseData

Hi support team,

I'm developing an application which uses PXCMFaceAnalysis.Landmark.PoseData to estimate a user's head pose, but I found all the parameters, yaw, pitch, roll were very inaccurate. For example, yaw angle of PoseData always returns between -5 to 5 degree no matter how I move my head. In my application, the accuracy of roll is not so important, but this SDK's yaw and pitch accuracy is not suitable.

So my requests are

Use multiple cameras

Hi, all

maybe this topic is not fresh, but it still bothers me recently. i'm trying to use two cameras(RGB+Depth) together.

I modified the sample code camera_uvmap and Example 92 (Streaming Data from Multiple Cameras) from SDK documentation

// streams is an array of PXCCapture::VideoStream instances from different cameras

PXCSmartArray<PXCImage> images(2);

PXCSmartSPArray sp(2);


// initialization

for (i=0;i<2;i++)



while (!stop) {

Intel-offered cam & official Senz3D difference


I'm investigating to use the Intel/Creative Depth Cam & the Perceptual Computing SDK for a project. I have seen the source engine demo at Cebit and was quite impressed.


However, I am confused which model to purchase:

Either the official Senz3D by Creative ( )

Subscribe to Intel® Perceptual Computing SDK