I have encountered this phenomenon a few times that after reconstruction, my dense cloud is quite spotty, and the points on the model are very unevenly distributed, with some neighboring areas having much more points than others, and some areas and “holes” with much less points in them. This can be seen on the example image I have attached. I am assuming this is maybe caused by how many features RC can extract from my images.
However, I was just wondering if this is caused by an error in my workflow, or maybe the quality of the images is not good enough? I was also wondering if this uneven distribution of points could have a negative effect on the quality of the model?
The model in the picture was reconstructed using default settings, no image downscaling for alignment, and 2x downscaling for reconstruction.
I’d appreciate if anyone could share thoughts on this.
It’s very hard to diagnose the issue by just looking at the point cloud. Please could you share as much information as possible about the project. I would like to see a 3d view where I can see all of the cameras pointing at this part of the mesh, and also the images from those cameras. It would be also good to see this same view in solid mode after texturing.
Thank you for your answer. The model is part of a forensic research project, and I am afraid I will not be able to post a screenshot showing the full model, nor will I be able to post any of the source images for privacy reasons.
However, I have attached a screenshot of approximately the same area in the solid view. You can see that those areas which appear less dense on the dense cloud have less detail in the solid view, however, in this view the difference is less obvious. Looking at the textured view, the difference is not visible at all.
I have also attached a screenshot showing the camera positions to give you an idea how the pictures were taken ( Following an elliptical path around the scanned object, taken from thee different heights, aswell as a few extra close-up shots to cover some areas which are a bit harder to see from the other angles ). Also, no images were left unaligned during the initial alignment process.
It might be that this phenomenon is normal and is just a result of how many features RC can pick out from images, but I was still wondering. If you can maybe suggest any settings that I could tweak and run the alignment and reconstruction again, please let me know.
I’m afraid that I will have to speculate a little. My guess would be that you have taken so few close-up photos. So the denser patches are where the few close-up cameras have found common features. You could probably get a consistently dense point cloud if you went around only 10 degrees at a time even on the close-up photos. Check out this video https://youtu.be/9e_NLp_FaUk