Post date: Nov 09, 2015 9:57:39 AM
After the preliminary bell models were completed, I went back to the fourth bell and masked the background out of all the photos. I reprocessed the model and got this result:
The question is whether this is significantly better than the older version, since the manual masking took about 3 hours for a person to sit and mask over 50 photographs. However, the bell tower model was much easier to mask since the blue sky allowed an automated tool to take much of the burden away and 30 minutes of labor resulted in this model:
Also to note, as I was reprocessing the model in a higher quality, the system got stuck in a state where over 37 GB of memory was used up between the RAM and pagefiles, so I had to abort the process and start over.