Kikuto

Hello , I read all the topics in SRWorks sections, but I didn't get any clue about how to solve this transformation problem. I also test SRHands example, but i had the same problem(left displacement) with depth images. Another related problem is the position of depth image in SRWorks samples in my device looks that it's in the same position of hand object. In SRWorks provided sample scene only the right eye can see the depth map occlusion, and the virtual camera was set on the left eye, so we can't see the depth occlusion by default in unity game windows. After making the proper adjustments, I get the same problem that depth mask is left displaced as when I saw in the right eye trough the headset, as virtual objects. I will post a photo of this problem. I belive that these problems are related. Thanks.

Hello everyone, Today I'm sharing with you my recent project in initial stage about the integration with Leap Motion Engine module. Please check my github repository: https://github.com/augustoicaro/HTC-Vive-Leap-Interaction-Engine-Integration/tree/master What is working: Hand models Interaction with UI Hover Grab (Partially) This project is a Working In Progress.

Hi Thanks for the fast reply. I tried your suggestion, but as the image below I still have 3D hand misplaced with my real hand in Hand Tracking SDK 0.8.1. So, how can I found the translation matrix that aligns these hands? Here is my unity example scene where the dependencies are SRWorks SDK 0.8.0.2 and Hand Track SDK 0.8.1. In this new SDK version, the hand detection is much better! Thanks for the hard work. Thanks.

With the SDK patch 0.8.0p1, now is possible to use Hand Tracking SDK + SRWorks in See Through mode. However, when I configure both SDKs together as in this unity example, I got different position for 3D hand and my real hand as in image bellow. This project was built on ViveSR/Sample/Sample scene. This problem occurs ins both hand in both eyes. how can I find the translation matrix that fixes this position? I intend to use handmat material, that made virtual hand invisible, to interact with virtual objects. As I finished the integration with Leap Motion Interaction Engine, I will be able to use all interactive features of Leap SDK. I will share the interaction code soon as I fix some small details. Thanks.

Hello , Thanks for sharing this patch. Now is possible to use handTrackSDK with SRWorks just configuring gesture provider in camera (eye) inside camera (eye offset) in any SRWorks see through example. But in my experiments, I experienced some initialization error popup that crashes unity, sometimes in my second run, sometimes in the first run, and now is working without problems. The latency is about 0.5 seconds with OpenCL using only the GPU that is good for what I want to do. Now I'm writing an interface code to use hand coordinates to create a leap motion hand and use the interaction engine. Best Regards.

Hi , I'm using 430.64, but as you said was a problem with unity version. Sorry, I didn't read that part in the plugin documentation. So I back to unity 2017.4.27f1 as in SRWorks project and everything works fine. Thanks again for your tip.

Yes, the latest version 0.8.0.2 for SRWorks and 0.8.0 for Hand Tracking SDK. Great news! Hand tracking in See Through will be a significant improvement :) In the path to use both, I had one problem with OpenCL error in Hand Tracking SDK for my GTX 1060 3gb, openCL just works in driver version 388.71, and I tried to put to work in newer versions, but no success. However, Depth map freezes in 388.71 driver version and just became functional in 390.77 driver version. I searched a lot and tried many things to solve this openCL problem. GPU-Z also checks OpenCL and CUDA. Had someone experienced this kind of issues?

Hello You forgot to answer the last question: Can I use HandSDK together with SRWorks in camera seethrough mode? I read the public code in HandSDK but there's no mention to camera input, so this part is in dll function that opens the camera and does steps of hand track algorithm. However, SRWorks are using cameras images, and when I try open camera again with HandSDK, I got a runtime error. So if there a way to pass the image of the frame to GetGestureResult() function avoiding get images from the camera will be possible to use HandSDK + SRWorks. So the point after GetGestureResult() accept the frame as input is what kind of image will be needed, left and right raw frame, after defish algorithm, or just one image.