Processing of a very large project. Your opinion / experience is needed!!!

[Vermesser]

Good evening!

A very large project is about to be processed.
I therefore ask myself a few questions about the right tactics and would be very interested in your experience and would be very grateful for any useful tips.

The whole project is very huge and I would divide it into several sub-areas. However, I reckon that even the partial areas could reach 50000 photos. 20 Mpixel JPGs are captured by a drone with a 4/3 CMOS sensor. These are urban areas. The desired result is the point cloud and the orthomosaic.

 I am now considering how many photos I can edit in Metashape 2.0 at the same time. What experiences have you had there? My PC has 128GB of RAM. Would it make sense to expand the working memory with an M2 SSD? Do I have to split the editing? But when? A common orientation of the photos with the highest possible quality would be desirable. After that, should I break down the editing across the region into sub-processes?

 I also considered processing on multiple servers, such as Amazon AWS. As I understand it, I then need a Pro version. That would incur enormous costs and I'm not really aware of the potential.

Sorry for the very long text!
I would appreciate your opinion and support!

Best regards
Patrick

[smiller]

Processing large photosets can be accomplished on a single PC/Mac/Linux workstation if there is sufficient RAM and the settings you used. Splitting them up into sub-chunks and then later aligning them to join can help sub-process to avoid potential crashes.

[ruyi7952]

You may consider renting https://www.agisoft.com/buy/saas/

The current version can support 50,000 images, which is stated in the manual.

For very large projects, the alignment step can be split into several sub-projects for processing. The rest of the steps, can be processed by clustering.