Just got a project that generated 12,000 images over a 500-acre site. My workstation (Ryzen 7, 32GB RAM, RTX 3070) is choking on this in Pix4D. Estimated processing time: 4 days.
Options I'm considering:
What do you guys do with massive datasets?
Sign in to join the discussion
Sign inSplit into sub-projects with 20-30% overlap between blocks. Process individually, then merge the outputs. This is the standard approach for large sites.
Make sure you have GCPs in the overlap areas so the blocks register properly. We routinely process 20,000+ image datasets this way.
128GB RAM would help but won't solve a 4-day processing time. The GPU is probably the bottleneck.
PIX4Dcloud handles large datasets much better than desktop processing. Upload your images and let their server farm handle it. Results in hours instead of days.
We switched all our large projects to cloud processing. The subscription cost is cheaper than the hardware upgrade would be.