I think I left the data in the original resolution but as jpeg when going from reality scan to brush. The reality is, that even with a very good lens, a 42mpx camera does not produce consistent 42mpx that are usable. The actual average effective resolution of the source-data is only 4k
Hi, I guess you are using RealityScan to export the colmap data later used in Brush ?
if that's the case, are you downsampling the original photos to 3840px to aligned them in RealityScan and they also be used in Brush in 3840px or are you doing the alignement in RealityScan with the Highres photos (8k maybe) and let Brush downsampling them to 3840px during the training ?
thanks !
I used the monopod to take photos in all four directions at a hight of ~1.2m. This was a rather dense grid with ~50cm distance between the points. (I first did all photos in one direction then in another and so on. So the grid was not consistent in all directions).
Then I extended the monopod to its full height and did the same with a sparser grid at that height. After that I took more photos angled down to the floor and points of interest and angled up to the ceiling.
I think this was about 500-600 photos. Yes, postshot can achieve that quality, but I would not rely on their alignment. For alignment Reality Scan or Metashape should be used.
In general I don't see the point in giving your money to the developers of postshot who sell the tech that others invented for free... rather use brush!
Hi, impressive sharpness and coverage !
I am curious to know how many photos did you took ?
Do you think this training could be achieve with Postshot ?