I use a Raspberry Pi to load and display my iPhoto library as a slideshow on a monitor hung in my den. Groovy! Except… a fair number of the images show up with strange rotation. Now, I’m thinking some of it might be EXIF stuff, which is why I’m here for Retrobatch (Fix EXIF Rotation), but I don’t think ALL of it is. 99.9% of the photos I’m displaying are just my kids doing stuff so I think it’d be easy (for some relative value of easy) for a ML filter to look at my photos and say “that’s landscape but should be portrait.” Is there such a thing already? Does this sound plausible?
The Fix EXIF Rotation node should do the trick. And yea - a ML filter would be cool to have for this but I don’t have any pointers for you right now on that.
Do you have a sample image where the Fix EXIF Rotation node doesn’t do the right thing for you?
I don’t, but I’m not saying this is a bug for you or for Apple. It’s just sometimes you’ll whip out the phone and it’ll take a second to reorient if you flip between portrait and landscape. Also, my iPhoto library is the family album so I get pictures from a bunch of different sources through different means, so I’m thinking some of them might not have proper orientation tags or were scanned paper pictures or the like. It’s a small thing, but the kind of thing it would seem would be pretty straightforward for ML. So straightforward I’m hoping someone has already done it and I don’t need to train something myself… but I guess my end use case (non-curated display of the whole library, bad or good photos) is kind of unusual.