When starting initial processing, PIX4Dmapper first computes keypoints on the images. It uses these keypoints to find matches between the images. The software runs an Automatic Aerial Triangulation (AAT) and Bundle Block Adjustment (BBA) from these initial matches.
If initial processing has already been done and is restarted, the software uses the existing keypoints and matches and starts the AAT and BBA. If the project type has changed from one processing to another, additional matches might be computed as the matching procedure is different for each project type.
Article feedback (for troubleshooting, post here instead)
I already processed a project but did not select one of the Point Cloud file exports that I need (XYZ). When I open Pix4d and reload the project do I have to re-run Step 1: Initial Processing or can I go straight to Step 2: Point Cloud?
In what scenarios does Step 1 need to be re-run?
You do not need to re-run any steps. If you click on raycloud<point clouds<densified point cloud<display properties on the left bar of the screen, you can export point cloud in the format you want. I am attaching a screenshot:
With larger projects, what we have been doing is running out initial processing, review our Quality report to make sure everything looks good, edit our check points, then run the whole process again (1. Initial Processing, 2. Point Cloud, 3. DSM, Orhomosaic). We are trying to save time, so I have a few questions regarding this process we have.
1. Once we are done with our check points, can we just reoptimize, then run steps 2. and 3. on the processing?
2. Ultimately we want our Keypoints image scale to be 1/2 (we have a 42MP camera). If we have to run the Initial processing twice, can we do rapid or 1/4 when we're just getting the Quality report and check points and then run 1/2 when processing 1, 2, and 3?
For 1, yes you can reoptimise and run step 2 and 3.
2. The reconstruction of the project depends on the image scale. The accuracy in turn depends on the reconstruction. Can you let me know why you are processing step 1 twice?