Notes:
- I don’t use F10. Reason is the number of images taken is maxed at about 20 (I now have the feeling this is linked to the number of slots available in Zwift’s internal thumbnail save.) Hence, I only initiate screen capture from the Companion App.
- All images taken are always available post-ride on both the PC and the phone (phone only has the HUD version, though.) None are corrupt or blank (@_Dean: Try looking in %HOMEPATH%\pictures\zwift…)
Here is the sequence of events, to the best of my understanding:
- Images are first saved to host PC disk (both “clean” and with HUD, up to a max of about 20 “clean” images., From that point on only HUD images are saved.
- The HUD version is then sent to the Companion, to be saved (using a different filename on the phone: 2020-08-15_10362769.jpg vs. 1597476975206.jpg, for example - the later being Linux millisecond epoch time, see converter here - why?)
- Upon completing a session, the user is presented with thumbnails of images to upload, to select form (I think this is done using some internal thumbnail cache.)
- Those images selected to upload are read from disk and sent to Zwift servers (hosted by Amazon) by the PC.
- As an image is uploaded, a note is added to the log. There is no mention of image upload for those cases where no thumbnail is presented (and no images are uploaded - but there are all there…)