Guys, why after spawning the object, when the camera is moved a little, is the object then repositioned?
I want to do a kind of virtual tour, and when I go inside the “object” and I turn around, the object is being repositioned. I do not want this behavior. I want the object to stay where I positioned it. (Second number 15)
I’m using studio.
Hi, welcome to the forum!
It looks like the SLAM tracking is losing sight of the ground and then attempting to re-position it back where it was. It’s hard to tell from the video since most of it is obscured by the model but I have a hunch that you’re tilting the camera up too much causing the Computer Vision algorithm to lose tracking of the floor.
The camera position was at 3/4. When it was 0 instead, it wasn’t tracking well. So I don’t know why it is doing that. I can try Y 2 but I don’t think it would change that much. Any solution? How can I share the code privately in the case?
Sure! Land your changes and share the project with the support workspace and I’ll take a look.
Thank you. I just shared. Please take a look and let me know. The file for instantiating is gridItem.ts
The tracking is looking pretty good to me. Keep in mind that the area I’m testing in is very visually rich with a lot of feature points for the World Tracking to hook onto. I also got good results in the Simulator as well.
Additionally, because 8th Wall experiences are responsive by default you will see a different scale depending on the initial Camera height, however that typically doesn’t affect tracking quality.
Yes, I may have been testing in an area with fewer feature points. But also when I try in an area with more feature points, sometimes it move the objects in another direction. I can’t understand why this happened.
You can see in the 36’.
In your video the floor is being tracked as the ground of the AR scene. The issue is that the real floor consists of repetitive wooden planks which all look similar to the engine when it is calculating the content’s location. This makes it harder for the system to maintain accurate tracking. If you test in an area with more unique visual features such as outside on concrete with cracks and distinct markings I would expect the tracking to perform much better. In the video I shared you can see that the floor is full of identifiable marks which helps the system lock on more reliably.