I want to mesh damaged vehicles for accident reconstruction purposes. When I mesh in pix4d, the mesh created is watertight, which causes some issues such as bubbles to appear and make the model visually unappealing. Also, I have ran the same project through Recap Photo, which seemed to produce a smoother mesh as well as a sharper image texture. In pix4d, the measuring tape markings can be seen but are blurry, but in the Recap Photo model, I can clearly read the tape markings. The image acquisition process was 3 cyclone type sets of photos at low, medium, and high relative elevations to the vehicle. My question is A) Is there a way to smooth out the noise in the mesh? B) How can I enhance the image texture quality C) Is there a way to make a mesh without being watertight as not to include fillers/bubbles? and D) What is the recommended picture count/acquisition procedure to acquire the best results for future projects?
Hi Kaye,
Would it be possible for you to share with us the project so we could better understand the issue? We understand that this may be a subject of confidentiality hence, in case of, please write us a request to support. In the request please include the link to this post so it would be easier for us to find the description here.
In case the project cannot be shared please send us the following:
- quality report
- p4d file
- log file
- screenshot of the mesh and point cloud.
Best,
Ina
Hi there,
I would be interested if there has been any progress/update on this request in regards mesh quality enhancement (avoiding bubbles, generation of a smoother and non-watertight mesh, reducing the level of noise). Has this request been resolved in the meantime; if so, would it please be possible to provide a quick update on how it could be better tackled?
Thank you and best regards,
Gabriel
Hello,
Version 4.5 and above generates cleaner point clouds, which means less manual cleaning and better 3D textured meshes.
See this blog for more information:
and feel free to get in touch with our sales team if you want to try it with your own dataset