Ransac to Mesh, whats best practice?
Posted: Sat Jan 28, 2017 12:11 am
First of all, Daniel, love your work and consistant work ethic and passion about your program, its the best Ive ever experianced. I could gush for pages.
I have a challanging modeling process. We have point clouds of electrical substations! The first time through I subsampled (bad idea), after that I used some of the other processes and got it going pretty well. (Exported out the model without ground, density filtered, subsectioned) Unfortunately after the mesh there was about 30 million faces! I simplified that to 1% to get the model to a workable fps, but by then after my botched rushed first learning attempt I had a day to sit down and look through the program to optimize the process.
Before I used poisson on subsections of the model with similair densitiez and then combined them after using the density filter on the mesh (bad normals, crap everywhere)
Now I looked at the Ransac result and its perfect in most regards (everything but the main transformer and the control house are cylinders!)
What would be a recommendation for replicating the ransac result, but as a mesh. I tried to use poisson, but I didnt have time to up the octree past 8. Ive been reading other forums and I think Ill be able to generate half decent normals now, but I cant seem to get an appropriately sized cylinder to mesh.
Any tips?
I have a challanging modeling process. We have point clouds of electrical substations! The first time through I subsampled (bad idea), after that I used some of the other processes and got it going pretty well. (Exported out the model without ground, density filtered, subsectioned) Unfortunately after the mesh there was about 30 million faces! I simplified that to 1% to get the model to a workable fps, but by then after my botched rushed first learning attempt I had a day to sit down and look through the program to optimize the process.
Before I used poisson on subsections of the model with similair densitiez and then combined them after using the density filter on the mesh (bad normals, crap everywhere)
Now I looked at the Ransac result and its perfect in most regards (everything but the main transformer and the control house are cylinders!)
What would be a recommendation for replicating the ransac result, but as a mesh. I tried to use poisson, but I didnt have time to up the octree past 8. Ive been reading other forums and I think Ill be able to generate half decent normals now, but I cant seem to get an appropriately sized cylinder to mesh.
Any tips?