Up to topic: Midjourney
We covered downloading your images from Midjourney for safekeeping or additional operations a little during one of the AI 101 sessions on Friday, 2024-01-26.
Since it's been a month or two, I decided to go get some downloaded, and I have to report, it's still very frustrating.
The process is pretty straightforward. Go to the website (can be regular, legacy, or alpha if you've got it, they seem to work pretty much the same right now), find your archive, then for each day, click a little "+" icon to select all the images for that day, then elsewhere on the screen, click the download button.
BE PATIENT; if you've got a lot of images for one day, it might take some seconds before anything happens (and you don't want to start two downloads for the same day, it'll just confuse you even more).
It will download all the images for that day, zipped into 50-image zip files, one zip file after another. HIGHLY RECOMMENDED: select the right download directory, then turn off the option, "Ask where to save each file before downloading" (or equivalent). You may have to "Allow multiple downloads for this site" one time, if something like that pops up.
Okay, the frustrating thing: for the most recent week, it uses relative day names, instead of dates (so, Today, Yesterday, 2 days ago, Wednesday, Tuesday, Monday), then on to dates (Jan 21, 2024). As far as I can tell, the day names are off by one, at least in my timezone (US Pacific).
It's helpful that the zip files have the dates of the images in the filename, HOWEVER, I just download one set of 2024-01-26_etc files for 2024-01-26, and another set named 2024-01-26_etc that were generated on 2024-01-25! Luckily, my browser wrote the second downloads with the same filename by appending "(1)" in the name, instead of overwriting them.
The legacy site, https://legacy.midjourney.com/ , used to support selecting ALL your images, across days, then downloading them in a bunch of zip files. It was super handy, although I'm sure there bugs that happened that way when there were a lot of files, so maybe it's better they're doing one day at a time.
I highly recommend keeping a spreadsheet or database when you're downloading, because you need to track what you've downloaded each time you do a backup. My columns are currently:
Date in interface, Date in zip filenames, Date downloaded, Number of images
Cross-checking the number of images reported in the web interface against the number of files in the zip filenames, and then when you unzip, helps make sure you've got all the files, and hopefully no duplicates; plus, it helps distinguish the days the images were generated, when the dates in the filenames aren't right.
We'll update this information as Midjourney continues to evolve -- hopefully for the better!