How do you manage Google Takeouts?

My google takeout is quite large as I’m sure most of yours are too. How do you manage them? Do you do a full takeout and merge the folders each time? Do you do the incremental one? Incremental seems confusing - does it go by date or by new files (eg, you add backdated photos). Since it has an end date you’d have to do a full takeout again to then incremental again to get back on track?)

There will hopefully be a time when I’m off of GPhotos but not anytime soon. Just wondering how people are doing this currently.

I cheat: I took a full google takeout a while ago, but from then on, I directly sync our phones’ original photos and videos to our NAS.

Note that you can create a “2020” album in Google Photos (by searching for 2020, clicking select on the most recent image, then scrolling down until lazy-loading finishes, then shift-clicking the select icon on the bottom-most image), and then do a Google Takeout that pulls down just that album.

I’d tried using a script to incrementally fetch my photos using the API, but Google strips several important tags from your photos and videos (including GPS metadata) when using their API, so it’s not really a solution.

I’d like to hear how others are handling this, too!

I “cheated,” too. I took a full Takeout and then switched my phones to sync new photos to the asset directory that Photostructure uses. Originally I deleted all the .json files form the Takeout directory, not knowing what they were. Photostructure worked fine, but just didn’t have any meta-data from Google. Recently I took another full Takeout and left the .json files alone, then forced Photostructure to Rebuild - it took days but did have metadata from Google, though not all metadata, strangely (e.g., it only had two “Faces” instead of the 10s or 100s that I defined in Google).

That’s no good: if you send me an example image and JSON sidecar that has more person tags, I can take a look at what’s going on.

I just realized that although I have 100s of faces tagged in Google Photos, only two of them have names assigned. Those are the 2 faces that Photostructure imported. So I’m thinking that if you don’t name your faces in Google Photos, the face metadata doesn’t get into the .json sidecar. Indeed, in the .json for a photo with a named-face is:

"people": [{
"name": "Joe Shmoe"

but the “people” object is missing from the .json of a photo with an un-named-face. :frowning: Oh well.

If you aren’t averse to giving Google even more data, you may want to spend a couple minutes adding those names and then do another Google Takeout: the new PhotoStructure face detection will leverage existing prior person names.

Juuust finished naming my faces in Google Photos. :slight_smile: Will do another Takeout as soon as I figure out my S3 issues.

I’m glad to hear you’re doing it exactly the same way I’m doing to prepare for (one day) moving away from Google Photos.

I also tried the script route, and realized some important metadata was lost on the way (most important to me is GPS metadata). This is what led me to find PhotoStructure in the first place.