It would be best if you remembered upload time to the server and costs for
server-side storage and bandwidth if you are looking at cloud server versus desktop. However, most cloud storage is approximately equal to hard drives in terms of cost when you get into the specifics, and link speeds continue to grow in line with evolving technology. There is just an increase in data transmission time, and the cloud offers the undeniable advantage of global availability.
More interestingly, for certain projects, the cloud provides the ability to
take the processing of point cloud to a completely different stage.
Advanced point cloud processing software can concurrently register several scans, and each performed on your CPU on a separate thread.
Up to 18 cores, or 36 threads, will pack cutting-edge hard drives. But the cloud will allow you to scale up this capability indefinitely. Theoretically, this allows the opportunity to perform the ‘coarse registration’ of a project of any scale simultaneously, exponentially decreasing registration times.
It will often take longer with more scans to place the scans in order setting scan pairs). You will keep adding threads in the cloud.
The need for automated registration
It is possible to see the opportunities and advantages of cloud computing for the development and processing of point clouds, but they still need to be realized.
Manual measures in conventional registration software at each level of alignment are one issue with supercharging the processing of point clouds in the ‘cloud.’ Accessing multi-thread processing is impractical, especially when using targetless scans manual inputs and cross-checks need to be handled through that process.
This is evolving as well. Multi-stage vector analysis software, for instance, enables coarse registration to be separated into clearly defined phases.
Rotational alignment can be accomplished independently with point clouds compressed into unique ‘vector spheres’ and quick 2D. This can speed up conventional hardware processing by 40 percent -80 percent. More critically, it provides a robust method for registering targetless scans, enabling front-loading of manual inputs. Entry to ‘cloud-scaled’ multi-thread processing is a fact through this automation. Survey teams can process scan sets of almost any size reasonably comparable amounts of time by combining this kind of processing software with cloud capabilities and wide bandwidth packages.
Looking to the future
The stumbling blocks to efficient point cloud formation have been point cloud processing and registration. Technology has always come down to this. The ability to align scans using natural features has been around for some time, and it has just never been sufficiently successful.