mf Thank you for your response. Actually, I noticed the abnormal orientation of the normal vectors on some faces and ran a script to fix them. However, the problem persists. When performing the Fuse operation, it seems to automatically readjust the orientation of the normal vectors, and the original problem reappears.
What's strange here is that although the Fuse operation is designed to adjust the direction of the normal vectors to ensure continuity across the shared edges of the faces (i think), it doesn't seem to work in my case. I suspect the problem may stem from some kind of geometric anomaly or topological defect at the intersections between the problematic faces and the normal faces. Despite my efforts, I have not yet been able to identify the exact cause and how to fix the error.
