Hello ppant,
There are several main factors that affect the processing time of the second phase of the dense cloud generation stage:
- CPU with low core frequency,
- excessive image overlap,
- ultra high quality used.
To fix the second case (excessive overlap) you can limit the number of the overlapping images taken into consideration, by using max_neighbors parameter in the buildDepthMaps and/or in the buildDenseCloud functions.
Since you already have the depth maps, you can try to run the dense cloud generation function based on the same depth maps using the mentioned parameter. For example, set to 50.
chunk.buildDenseCloud(max_neighbors = 50)
However, you can also consider using lower quality options for the aerial datasets.