Just Installed Reality Scan 2 and did a little slow walking 8K 360 video of an office interior. Slowly walked around the office in a circle.
The i imported the video sequence and clicked (F5) create a complete 3D reconstruction with texture and vertex colors.
My pc is Intel 285K with RTX5090 GPU and Gen5 2TB SSD. The camera i used i an insta360 x5 8k.
First issue is that the result is a complete mess ( see attached screenshot). Am i doing something wrong? Any special settings i need to adjust? Is it because i was holding the camera above my head?
During the import and the rendering my CPU was at almost 100% utilization, while my RTX5090 GPU was at like 2%. Are there more settings for that as well that i need to change?
Hello @iTosaf
First of all, RealityScan doesn’t directly support 360 images/videos. Also, too distorted data could be problematic. The second problem here could be usage of the video, as from the video you’ll get worse model as just from images. Another issue could be number of used images (as you used only 61), the capturing path (rooms require specific capturing path), overlap between the images and also captured object (as it need to have features for proper alignment, on your images there is window, shiny objects and featureless parts of the room).
For cases like rooms the combination of laser scans and photogrammetry is most optimal option. But I understand that it is not always possible.
In this case I wouldn’t be using video or 360, I would change the image’s overlap and amount and follow the “photogrammetric” path to capture the room.
In this tutorial is mentioned also how to capture interiors: https://www.youtube.com/watch?v=9e_NLp_FaUk
