Pixiv IDs by year

Recently decided to roughly divide Pixiv IDs by year so that it's easier to find things uploaded at a later date but from that period.
Opening this forum as I'm not entirely sure where to put this information as it feels out of place to slap into a wiki (considered pixiv but still feels weird there). Feel free to suggest a place to put these if you find them useful. At the very least, this forum will serve as a place to find it if people really want it.

There's a couple of improvements I know I can make such as being closer to midnight on the 1st of January and also making the timezone it's based on either UTC+0 or UTC+9 (currently UTC+1 because I'm in British Summer Time right now).

The search can be used freely for those with limited tag searches. Can also easily just take the start of one year and the end of another for a more broader style search (2013-2015 for example pixiv:32538706..54397931). I do recommend adding -dairi right now though, sorry..

(Edited to correct and simplify the searches as per BrokenEagle's suggestion. Thanks!! ^^)

I wonder if it's possible for this to also be done with other popular sources on Danbooru, in particular Twitter.

You know, as someone prone to ending up in strange places on Pixiv and Twitter (not to mention knowing how popular it is still to upload images from 2013 and earlier), I think having tags based off these systems would be kinda cool (though it'd probably be annoying to upkeep).

But in any case, being able to search this at all is really fucking cool. Thanks for placing them somewhere to show us man.

on a similar note, would love to be able to search by the source's upload date rather than ours by means of some sourceage: or sourcedate: metatag. some of the works on our site are uploaded months, sometimes years after the original post, as evident by looking at the earlier years on these searches.

the difficult part would of course be backfilling this data from existing posts, and deciding what to do with bad ids (maybe take the date of the previous and next available id and average it out?)

Idk how feasible it is but I do know the date and time of tweets are extractable. If it doesn't strain the site in some way then just pixiv and twitter alone would cover most of the site (8m~ I think) for a theoretical sourceage meta tag

Kaleidoscoped said:

the difficult part would of course be backfilling this data from existing posts, and deciding what to do with bad ids (maybe take the date of the previous and next available id and average it out?)

Depending on tools developed, you could probably also leverage archived sources like the Wayback Machine or archive.is.

Calculating the post time of a Tweet is possible even in the case of bad id posts. All you need is the ID itself.
Someone already has done this work thankfully:

https://github.com/oduwsdl/tweetedat
https://oduwsdl.github.io/tweetedat/

The only thing to keep in mind is that OUR data needs to be correct as well.
The oldest Twitter sourced image at the moment is post #2963 which was uploaded nearly 20 years ago.
But its source is currently set to a Tweet from 2018 which obviously will return wrong data.

Note that this does not work for any Tweets that were made before Snowflake was implemented, somewhere around late 2010. Older Tweets used a different system that does not embed time data.

The only thing to keep in mind is that OUR data needs to be correct as well.
The oldest Twitter sourced image at the moment is post #2963 which was uploaded nearly 20 years ago.
But its source is currently set to a Tweet from 2018 which obviously will return wrong data.

Removed the source from that post because it's wrong. The source is supposed to be "where the image was originally found" not "where an inexact repost was found 20 years later".

The point stands though that sources can be wrong and when the image was posted and when the image was originally created are two different things. It's possible for an artist to draw something in 2005, then post it on Pixiv in 2010, then repost it on Twitter multiple times starting in 2020, leading to the same image being associated with multiple dates. This is why "just store the date when it was created" is not as straightforward as you might think.

evazion said:

Removed the source from that post because it's wrong. The source is supposed to be "where the image was originally found" not "where an inexact repost was found 20 years later".

Since the commentary wasn't removed, I readded the source as a TN in the commentary. Hopefully, if the actual original source is found, the commentary can be swapped out for whatever the original source had.

The point stands though that sources can be wrong and when the image was posted and when the image was originally created are two different things. It's possible for an artist to draw something in 2005, then post it on Pixiv in 2010, then repost it on Twitter multiple times starting in 2020, leading to the same image being associated with multiple dates. This is why "just store the date when it was created" is not as straightforward as you might think.

Yeah, that's always possible, but I think that's less of an issue than people putting inexact MD5 mismatching reposts as the source for an older post.

thankyou for this valuable information, this thread is so meta...

Thanks for this! It's really neat seeing how art from pixiv has evolved over the years.

I support having sourcedate metadata. I think it's more valuable than post date.
In the case of same image conflicts, maybe it's possible to just use the date of the source post or generate a table in the comments for the different dates.

Just thought I'd chime in, but those pixiv: metatags aren't doing what you think they're doing. In each case, you're using the greater than ">" or less than "<" symbols, which isn't inclusive of the number that appends it. You would want to use greater than or equals ">=" or less than or equals "<=" to be inclusive. As it is right now, you're not including around 19 Pixiv IDs.

Also, you can simplify all of those metatags by using range syntax instead. You can either use the double dot ".." which is inclusive on both sides (pixiv:1..294842, i.e. greater than equal to 1 and less than or equal to 294842), or the triple dot "..." which is only inclusive on the left side (pixiv:1...294842, i.e. greater than equal to 1 and less than 294842).

Check out Help:Cheatsheet#Range qualifiers for more information.