Support Website Contact Support Blog

Does cloud processing use full or half image size

Hello all, I would like to buy the licence for Pix 4d and because I dont have a good PC, I would like to benefit from the cloud processing feature. However, by trying the free 15day desktop app, I noticed that on the presets, it only uses half the image size. Does that affect the quality and resolution of the output and hence the Ground Sampling Distance?

Thank you!

Hello Pavlos,

Different processing templates are applied to the project according to how this is uploaded to the Cloud platform. A project can be uploaded from:

  • The Pix4Dcapture app
  • The Cloud web platform
  • The desktop application

You can read a full description of the cases and the corresponding applied templates here.

If you are creating a project directly from the Cloud platform, a 3D Maps template is applied that, as you mentioned, uses 1/2 image scale for the point cloud densification .

The processing options are customizable on the Cloud, provided that the project is created on the desktop application, the custom processing options chosen and the project finally uploaded to the Cloud for processing. More information in: How to edit the options for processing on the Cloud.

Regarding your question Does that [half image scale for the point cloud densification] affect the quality and resolution of the output and hence the Ground Sampling Distance? , the GSD only depends on the focal length, flight height, image and sensor width. See our GSD calculator. The average GSD value is calculated during Step 1 so the image scale of Step 2 processing options is not influencing this number in any form. Keep in mind that the GSD written in the quality report and used for the 2D outputs is an average value due to terrain elevation differences and changes in the angle of the camera while shooting.

The half image size is, based on our experience, the recommended value. Using the original image size, computes more points than the half-scale but this option is slower and may require four times more RAM. Usually it does not significantly improve the results.

Notice that fewer points in the point cloud will generate a less accurate orthomosaic, as the DSM will be more interpolated especially in areas where the terrain changes a lot or where there are a lot of details. In general, the relative accuracy of the outputs depends on many factors, among which the overlap and the visual content.

I hope that helps :slight_smile:

I can’t really thank you enough for such a detailed answer! It really cleared things up in my head! I also found really interesting that i can import the GCPs before processing in the cloud. I believe that you still have to carry out the first step of processing on the PC at first.

I hope it helps others too!!
Thanks again Alice!!!

Pavlos,

I am happy I was helpful!

To add GCPs on the Cloud you don’t need to process locally. The only thing you need to do is adding the GCPs and mark them. Here the Support article explaining the workflow, just in case others need it.

Cheers

Thank you so much again! I hadn’t realized that you can add them at this step!!