I’m a long-time user of Reality Capture and now Reality Scan, and I’m running into a ton of really terrible bugs with the latest version. It’s so bad that it’s unusable.
Here are examples:
The texturing looks like absolute garbage, regardless of settings. This is part of a model made with 3000 20mp images taken 12 ft from the building. I’ve used the default, my (previous) preferred settings, multiple different combinations, and this is the best one. It literally looks like a texture isn’t applied.
it looks like your project is somehow corrupted. According to the attached ortho, its size is too big. Is your project georeferenced? If so, how did it look after first alignment? Do you have some camera pose residuals? Are the images placed properly in the map view? Have you used control points or GCPs? What kind of drone have you used? The texture you are showing, is it in 3D view?
These are two different projects and datasets. The ortho/texture project isn’t georeferenced, because it was down between skyscrapers and the RTK data was unreliable at best. I didn’t use control points, but I would normally just scale based on known measurements.
I built the model and shot the ortho using the reconstruction region and “true ortho on textured model” and that was the angle that it produced. The size is arbitrary, but I don’t recall seeing it listed as meters when all positioning is “off.”
The crop circles project is another similar building. The dataset worked in metashape after I encountered these problems.
I’m going uninstall and reinstall and see if I can replicate the issues. Another colleague of mine is reporting similar texture problems - it looks like “flakes” for want of a better descriptor.
As there are feet for the ortho it needs to be georeferenced or at least scaled. When the project is not georeferenced instead of ft there is units. The size is too extreme and I suppose it somehow confused the application.
After reinstall please provide also the application reset (make it like a clean install workflow). If you will get the same results, would it be possible to share your dataset with us?
