The “Align Photos” step produces a sparse point cloud. This is the first output and the first ”check” that the model must pass before proceeding.
Aligning Photos Workflow #
- First, go to Workflow> Align Photos and use the following parameters:
| General | |
|---|---|
| Accuracy | Medium |
| Generic preselection | Checked |
| Reference preselection | Unchecked/Source |
| Reset current alignment | Unchecked |
| Advanced | |
|---|---|
| Key point limit | 10,000 |
| Tie point limit: | 1,000 |
| Apply Masks To | |
|---|---|
| Exclude stationary tie points | Unchecked |
| Guided image matchmaking | Unchecked |
| Adaptive camera model fitting | Unchecked |
- If you are using photos taken by the drone, make sure that Reference preselection is unchecked before aligning the photos.
- After the photos are aligned, the program will generate a sparse point cloud. Check the sparse cloud from a number of angles to see if there are any obvious gaps, holes, or areas where the points seem too sparse.
If you judge that the sparse cloud is acceptable…
…proceed by moving to the next step.
If you judge that the sparse could is NOT acceptable…
- Check that all the photos were aligned.
- Click the camera icon to turn on all the cameras. Then, use the Rectangle Selection tool to select all visible cameras. If all the photos in the photos window are not selected, you may be able to fill gaps by realigning the photos.
- To do this, select all the photos in the photo window, then right-click and select Align Selected Cameras. If the gaps are filled in the resulting sparse cloud, you are ready to move on.
- If the photos are not aligned even after realignment, try to align photos with higher key-point and tie-point limits. Recommended limits are as follows: Key points: 40,000; Tie points: 3,000.
- If all photos were aligned already, let the excavator know where the gaps appear in the model and request additional photos. At this point, you should return to the “Add Photos” step in the main workflow and start the process again with the additional photos.
Note: An important part of the modeling process is being aware of where the context begins and ends (in other words, what parts of the model are more important than others). Usually, the area within the CTs is the context; therefore, any area outside of the CTs is less important, and may contain gaps/distortions without causing problems. However, if a context is against a scarp or a wall, sometimes the CTs will be far away from the actual context in order to surround it on all sides. If you model a context where one or more CTs appear to be far away from the others and there is a large gap, make sure to communicate with the excavators about whether or not the gap might represent a wall or a scarp that is not part of the context.