I’ve just started making digital elevation models using photogrammetry from drone photos.
I’ve used mostly the default pipeline in the latest version of Meshroom to create a textured mesh from 700 photos. I say mostly default as I’ve bypassed depth mapping for now and made a mesh straight from structure for motion. Depth mapping takes an age!
Once I import the OBJ file into Blender, the texture is inside the terrain model. At this stage the model is not closed, so you flip it too see the high quality texturing on the wrong side. I can see a poor quality version through the “right”side.
Everything else is great, coherent texture and detailed, sensible mesh.
What I’ve tried:
Only control N recalculate normals which didn’t work
Thanks for looking.
This may help. I’ve attached a image of the screen from Meshroom. It is the structure from motion view. Colored point cloud of the landscape. I’ve annotated with white arrow to show Y axis. The side of the 3D model the photos should end up on is pointed to with a yellow arrow.this is where all the camera views are reconstructed in mesh room
Maybe something to do with the weird orientation in XYZ space and a transformation is needed?
Second edit! Didn’t answer the question yet as I have one last thing to figure out. Turns out and X and Y transform on the textured mesh put the texture on the right side. But the zoom (tracking??) still won’t let me zoom in on the detail on the right side