This is a wrapper around PyVips and Vips to enable me to batch convert images from Heic to Jpg for reporting.
Project has been updated to utilise pillow_heif
Currently the script can process approximately 13.5 photos per second, depending on available cores, as it uses the concurrent.futures module.. See the table below for the speed comparison before and after adding multithreading.
More information and quick tutorial on concurrent.futures: https://gist.github.com/mangecoeur/9540178
| Number of Photos | Elapsed Time | Seconds/photo | |
|---|---|---|---|
| VIPS No Multithreading | 9 | 21.97 seconds | 2.44 |
| VIPS Mulithreading with concurrent.futures | 9 | 8.64 seconds | 0.96 |
| Pillow and Pillow_heif Multithreading with concurrent.futures | 9 | 4.38 seconds | 0.49 |
Having changed computer since, I didn't want to rerun the above numbers to get fair results, but starting with the last item, further improvements are shown below:
| Number of Photos | Elapsed Time | Seconds/photo | |
|---|---|---|---|
| Pillow and Pillow_heif Multithreading with concurrent.futures | 120 | 26.9 seconds | 0.224 |
| Multithreading with chunksize = 10 | 120 | 20.1 seconds | 0.168 |
| Multithreading with chunksize = 5 | 120 | 19.32 seconds | 0.161 |
| Multithreading with chunksize = 3 & directory traversing improvement | 120 | 8.9 seconds | 0.074 |