When I run the Meshing in the Magic Leap 2 Example, it generates a mesh of the space, but only about 40% of the entire room. Compared to Magic Leap 1, the meshing in Magic Leap 2 seems to be much less accurate.
The mesh is not generated, especially for the floor.
Both spatial mapping and custom fitting have been performed.
Please let me know if there is any way to improve the accuracy of the mesh of the space.
Can anyone answer about this meshing? I feel that this accuracy is indeed not useful. Not only the meshing but also the Plane was not very accurate. Any suggestions for a solution, such as the possibility that the spatial mapping is not working properly? Is it just in my environment and is meshing working properly for others?
I am not in a great hurry to answer.
However, unanswered as it is is quite troubling.
The accuracy of the current meshing is, in my opinion, of a quality that I personally can't stand in practice.
This is a problem that depends on the environment of my room, and in most places this If the event does not occur, under what circumstances is it difficult to generate a mesh?
Please let me know if there are any prerequisites for performing meshing.
I have had great success in utilizing the meshing example in the ML2. I even extracted it to a separate project and utilize it in an application I am working on. I have been quite satisfied with its current functionality/potential. I have noticed some minor gaps here and there but considering everything like lighting, reflective surfaces, windows...I would like to say it performs very well. I have had success with getting a full 20x 20 room to display with what I would describe as nearly 100% correct mapping besides some glass window panes in which case it maps the space behind the glass correctly.
There are many settings that you can change when utilizing the Meshing example I would double check those settings and maybe mess around with those setting to see if that has an effect on your current results.
Just for reference the image shows the settings I use for my results. I believe these are mostly default settings from the ML2 meshing example.
edited Also attached some captured results on the ML2 device. You can ignore the Red mesh if you like.(although it is still mapping correctly) I have made some modifications and added some additional content to the default meshing example. As you can see the floors, walls and everything in my office appears to be scanned and mapped correctly.
@robert.ellis Thanks for the advice.
I will change the settings and verify the information you gave us. @sengelman The information I can give you is current, but the sticker has not been removed yet. I will check the dirt on the lens later. Eye calibration has been performed. The settings are as in the provided Example and have not been touched in any way.
After this posting, the meshing has not been re-verified yet. I would like to conduct another verification based on the responses I have received from you guys.
Thank you, @tokufxug . If the sticker is on the depth sensor, that's what I suspect is giving you these issues. Thanks for re-verifying; 'looking forward to working through this more.
@sengelman Once the protective sheet was removed, the accuracy of mesh generation improved.
The mesh generation of Magic Leap 2 seemed faster than that of Magic Leap 1. The mesh is also generated for the two staircases, and the mesh is also generated when moving from one room to another. It seems to be difficult to generate meshes for black objects.
