I am trying to process forestry data, but the outputs are vert poor quality.
And here is processing summary:
Is there anything I can do to improve the results?
Settings:
auto-boundary:true, fast-orthophoto:true, feature-quality:ultra, matcher-type:bruteforce, min-num-features:20000, optimize-disk-space:true, pc-quality:lowest, radiometric-calibration:camera
Forests are hard.
Overlap on forests should probably be 77-83% for best results, and flight height for that overlap calculation needs to be calculated at the height of the trees, not the height of the ground.
auto-boundary:true, fast-orthophoto:true, feature-quality:ultra, matcher-type:bruteforce, min-num-features:20000, optimize-disk-space:true, pc-quality:lowest, radiometric-calibration:camera
I would start here. 
You might also go with default feature-quality. Ultra isn’t always better.
If you’re able to share data, sometimes folks are kind enough to run your dataset through and figure out good customized settings.
thanks! Unfortunately, I wasn’t the one who collected the data and side overlap is very weak… I am just trying to get the best possible results with what I have. Would you mind to explain why ultra is not the best, I thought ultra->more features->easier to match, am I wrong in this logic?
You also suggested processing with regular point cloud quality, how does point cloud affect orthophoto? Would going with higher pc quality help?
Appreciate your help!
Yes, as far as I know from the calculated / processed pointcloud the 2D orthophoto is generated. So a ‘good’ pointcloud results in a ‘good’ orthophoto.
As mentioned above you could try with feature-quality:high and pc-quality:high. Forests are always a bit tricky.
