Please tell me your recommended PC specs for Context Capture

Latest Nvidia gaming card and much RAM at least 64GB or 128GB. For faster production you would need CC center edition license to run on multiple machines.

Maybe try Contextcapture Console(ContextCapture Cloud Processing Service) you could ask for trial run.

A single engine on a high-end desktop will reconstruct between 20-30 Gpix per 24h and around 10-15 Gpix on the cloud. The cloud can engage up to 10 engines.

Thank you for your response. My question was not too realistic. .

Do you know how long it will take for 3d modeling with the following PC specs?

PC1

CPU:Core i7 8750H

RAM:16GB

GPU:NVIDIA Geforce GTX 1060 6GB

PC2

CPU:Core i7 8750H

RAM:32GB

GPU:NVIDIA Geforce GTX 1080 Ti 11GB

PC3

CPU:Core i7 8750H

RAM:16

GPU:NVIDIA Quadro P3200 6GB GDDR5

For us single machine 16Gpix with 64GB requires split in 3 tiles with each tile requiring 8hours so around 1Gpix in hour. You have around 25Gpix of photos which will took 24hours or less on one machine.

3 machines could give 8hour processing each part of model depending on tile count. Quadro P3200 is too slow for CC so you actually have only 2 decent machines.

Hey Oto, just out of curiosity .. what was your RAM estimate for that? We've always had guidance to aim for 16GB or less for the estimate regardless of what was in the actual box. In fact we can VM's on our virtual platform with anything up to ~ 300GB .. 396 GB in the physical host so can comfortably get up here. But watching the performance monitors - we get terribly inefficient processing. Small tiles seem much slower than big tiles.

And assuming you are running bigger tiles .. do you run the full LOD's? So across tile LOD? We have had problems with LOD generation and running out of "RAM" - but best I figured it was a ratio between RAM and the paging disk size .. need something like 4:1 .. disk to RAM, otherwise the LOD is trying to allocate too much "memory".

The worst part .. when you have job with thousands of tiles .. if a tile fails it can be tough to get it out when we are already near the limits of the box.

Hey Oto, just out of curiosity .. what was your RAM estimate for that? We've always had guidance to aim for 16GB or less for the estimate regardless of what was in the actual box. In fact we can VM's on our virtual platform with anything up to ~ 300GB .. 396 GB in the physical host so can comfortably get up here. But watching the performance monitors - we get terribly inefficient processing. Small tiles seem much slower than big tiles.

And assuming you are running bigger tiles .. do you run the full LOD's? So across tile LOD? We have had problems with LOD generation and running out of "RAM" - but best I figured it was a ratio between RAM and the paging disk size .. need something like 4:1 .. disk to RAM, otherwise the LOD is trying to allocate too much "memory".

The worst part .. when you have job with thousands of tiles .. if a tile fails it can be tough to get it out when we are already near the limits of the box.