I am scanning artwork from 20mm (point distance 0,01mm) up to 1,5 m (point distance 0,05 mm) - often 50mio Polys.
What scanner hardware and software are you currently using? What are the dimensions of the physical objects you’re scanning and how many points are in the cloud data (scan data) and then how many points in the converted stl data? Are you pulling color info or just the physical scan data?
The task is to get a 3D Model as original as possible to machine things like skin, folds, seams… as a perfect copy to create a 3D CAD.
Do you need CAD or STL data as your final output? There is a big difference and is the primary one that determines the workflow here.
Often I have to modify the global shape without loosing details and - if possible I have to work on original STL - no remesh or zremesher and project and so on.
No you don’t have to work on the original STL. You can remesh (Zremesh) with a projection (and I usually recommend it). I do this all the time with scan data and my clients cannot tell the difference between the stl and remeshed Zbrush file. I get all the original data with the benefits of Zbrush tools for modification/detailing.
What I am looking for is a solution to import such huge files and modify them in ZB (3, 4 better 5GB) - if it´s not possible with ZB64bit I have to look for another soft - for example Geomagic Freeform.
Geomagic Freeform isn’t going to give you what you’re looking for here. And that’s an expensive tool that will probably cost you a minimum of $5,000. No offense but what you’re looking for may be the wrong idea. The thing you should be looking for is an overall proper workflow. If your scanning hardware/software isn’t set up right it could be gathering redundant information and easily bloating your file size.
If I could look at your scan data file (stl) I could probably give you better information that would help and point you in the right direction.