The Mixed Reality Forums here are no longer being used or maintained.
There are a few other places we would like to direct you to for support, both from Microsoft and from the community.
The first way we want to connect with you is our mixed reality developer program, which you can sign up for at https://aka.ms/IWantMR.
For technical questions, please use Stack Overflow, and tag your questions using either hololens or windows-mixed-reality.
If you want to join in discussions, please do so in the HoloDevelopers Slack, which you can join by going to https://aka.ms/holodevelopers, or in our Microsoft Tech Communities forums at https://techcommunity.microsoft.com/t5/mixed-reality/ct-p/MicrosoftMixedReality.
And always feel free to hit us up on Twitter @MxdRealityDev.
Visible shifting of holograms relative to real-world objects?
Hi,
I am attempting to do precise placement of virtual objects on top of real-world objects. However, I am encountering some severe shifting of any virtual objects relative to the location of the real-world objects they should be anchored to.
Here is a video (low-res and blurry, sorry) illustrating my issue:
https://youtu.be/G4g0n9gK-NM
To explain what's going on:
- I am using the provided SpatialProcessing script to generate a series of planes from the Spatial Mesh of my environment.
- Because the generated planes may not exactly coincide with a real-life plane, I have set up my application to manually adjust the location of these planes (for example, so the floor plane is actually at the same world-space height as the real floor).
- To assist with this, I've set up a texture on the planes (the grid you're seeing) that is scaled based on real-world units. So the space between each red line is 10 cm (confirmed with calipers).
- Each plane, upon creation, has a World Anchor script attached to it.
- I am setting the stabilization plane each frame to be based on the user-to-world raycast location. In addition, the stabilization plane is being placed so that its normal matches the normal of the raycast hit. The only objects that are being hit by this raycast are the planes (not the underlying coarse-geometry). This means that as I move my head around the scene, as long as I am gazing at the same plane, the stabilization plane itself is not varying (because the normal is the same).
As you can see, the issue is that objects in the real world appear to shift relative to the virtual grid. Note the marker placed on the floor, and how the virtual grid lines on the floor plane don't remain anchored relative to it. This is visible both in the mixed-reality capture and when I am wearing the headset. The shift is much larger than what I was expecting (>2-3 cm) and what is frustrating is that the shift varies as my head moves relative to the marker.
Is there anything I am missing in terms of improving the alignment/anchoring? Has anyone been able to get virtual objects to appear as if they are rigidly anchored in the real world?
Answers
Hi @DanAndersen,
From the video, it looks like the floor is actually reflective, which will cause problems for the sensors. Just for comparison, have you test this out on a carpeted or otherwise non-reflective floor?
James Ashley
VS 2017 v5.3.3, Unity 2017.3.0f3, MRTK 2017.1.2, W10 17063
Microsoft MVP, Freelance HoloLens/MR Developer
www.imaginativeuniversal.com
Here is a recording with the objects on a carpeted floor:
https://youtu.be/zHiJVwBU8ZI
Since my last recording I've also gone through to ensure that my rendering is at 60 FPS (in the recording it of course is locked to 30 FPS). The misalignment in the recording is still ~2-3 cm. When I ensure 60 FPS rendering, when it's not recording, the alignment is better but not as good as expected (still drifting around 1-2 cm).
Any follow up solutions to improving stability relative to real world objects?
Unfortunately, I hadn't found a good solution to this issue.
Hello @DanAndersen, Did you improve this in any way?
Hi,
I haven't looked into this recently, so I can't say for sure. Recent HoloLens SDKs might have made improvements, but I don't know (the research I'm currently doing doesn't involve alignment against real-world objects to that level of precision).