Touhou Pixiv tag research

[Tenkko]

I spent some time searching the different Windows era characters' tags on Pixiv, over different time periods, and compiled the results. I still need some time to finish the spreadsheet but I got a few graphs in the meantime.

Last one is for characters and is very long

[Isaac]

This research was interesting to see, especially for the characters and the general touhou tags. Thanks for conducting it!

[Tenkko]

It kinda came from me wanting to see just how skewed the spread was. Turns out very! But at the same time there really haven't been any big hitters since SA huh. Well not in Pixiv at least.

There's a bunch of interesting comparisons to draw from the rest of the data too, I'll post my findings once I finish organizing that.

[TheM3ds]

32 minutes ago, Tenkko said:

It kinda came from me wanting to see just how skewed the spread was. Turns out very! But at the same time there really haven't been any big hitters since SA huh. Well not in Pixiv at least.

There's a bunch of interesting comparisons to draw from the rest of the data too, I'll post my findings once I finish organizing that.

Yeah, I wanted to point out how interesting the statistics are, but seeing characters like unzan and yorihime so low is funny to me. Still it's nice seeing characters like Kokoro and Koishi getting some attention.

Edit: My brain just doesn't want to work right now, huh?

[Isaac]

21 minutes ago, Tenkko said:

It kinda came from me wanting to see just how skewed the spread was. Turns out very! But at the same time there really haven't been any big hitters since SA huh. Well not in Pixiv at least.

There's a bunch of interesting comparisons to draw from the rest of the data too, I'll post my findings once I finish organizing that.

Yeah, SA is my favourite one and the cast as well as music is awesome. Satori and Koishi are within the top 10 most popular too.

Looking forward to the rest of the data!

[Isaac]

1 hour ago, TheM3ds said:

Yeah, I wanted to point out how interesting the statistics are, but seeing characters like unzan and yorihime so low is funny to me. Still it's nice seeing characters like Kokoro and Koishi getting some attention.

Another thing I love about touhou is that every character seems to have a fanbase, even if they are less popular. There are some I follow on twitter and pixiv whose favourite is Yorihime which is great to see! My personal favourite is Parsee!

[Tenkko]

Alright, here's the link to the spreadsheet. Note that all numbers are approximations yada yada

Some of my observations:

• EoSD consistently holds about 20% of the Touhou tag
• HSiFS has the lowest numbers overall
• IN was the 3rd most popular cast until around 2009 when MoF took off
• Similarly MoF was passed by SA around 2015, but they’re still about the same
• TD, DDC, LoLK and WBaWC are around the same place
• PoFV does surprisingly well despite having less characters (because of Aya)
• UM’s doing good numbers so far

onto the characters

• Marisa had (barely) more hits than Reimu until 2009
• Rumia had a big jump in 2010 and kept going higher, peaking in 2016
• Remilia got more yearly hits than her sister until 2016. something similar happened with Satori and Koishi
• Renko > Maribel
• Tewi had high numbers until a sharp drop in 2012, which is also when Reisen began her climb
• Surprisingly Eirin almost consistently has the lowest numbers of the IN cast
• Mokou > Kaguya. by far.
• Sunny and Luna had similar numbers for a while, while Star had the lowest count. this changed in 2019 with Star passing them both.
• If not Aya then Yuuka is at the top of PoFV, followed by Eiki and then Komachi
• Medicine is the lowest in PoFV but does relatively well overall
• Sanae is MoF’s best, often followed by Momiji
• Meanwhile Kanako is among the lowest. on that note, Suwako > Kanako
• Tenshi has the highest numbers of the spinoff characters
• Satori got way more yearly hits than Koishi until 2011, and since 2018 Koishi consistently has around half of SA’s hits per year
• Nazrin was very popular for a long while it seems
• Futo gets the most hits from TD, closely followed by Miko. meanwhile Mamizou remains at the very bottom surprisingly
• Of the Grassroots youkai, Kagerou quickly took the highest place followed by Banki and then Hime
• LoLK is a battle between Junko and Piece
• Sagume also gets a lot of attention huh
• Doremy had the most out of LoLK in 2018 (AoCF influence?)
• Sumireko had a small jump in 2018 too (VD?)
• Aunn and Okina quickly became the (hidden) stars of HSiFS
• Narumi had a huge jump in 2019. went right back down immediately after though
• Shion > Joon. though overall both drop off after their first year
• Mayumi had high numbers first two years, but dropped in 2021
• Yachie is on the rise
• About UM, Tsukasa got the most hits until the end of 2021, followed by Megumu
• On the lower end there’s Sannyo and Misumaru
• Yuuma is doing well

There's also some comparisons with the popularity polls but there's little correlation. And while I didn't count PC98, going by the numbers it should be about 15-18% of the Touhou tag (along with everything else I didn't search)

[Isaac]

2 hours ago, Tenkko said:

Alright, here's the link to the spreadsheet. Note that all numbers are approximations yada yada

Some of my observations:

• EoSD consistently holds about 20% of the Touhou tag
• HSiFS has the lowest numbers overall
• IN was the 3rd most popular cast until around 2009 when MoF took off
• Similarly MoF was passed by SA around 2015, but they’re still about the same
• TD, DDC, LoLK and WBaWC are around the same place
• PoFV does surprisingly well despite having less characters (because of Aya)
• UM’s doing good numbers so far

onto the characters

• Marisa had (barely) more hits than Reimu until 2009
• Rumia had a big jump in 2010 and kept going higher, peaking in 2016
• Remilia got more yearly hits than her sister until 2016. something similar happened with Satori and Koishi
• Renko > Maribel
• Tewi had high numbers until a sharp drop in 2012, which is also when Reisen began her climb
• Surprisingly Eirin almost consistently has the lowest numbers of the IN cast
• Mokou > Kaguya. by far.
• Sunny and Luna had similar numbers for a while, while Star had the lowest count. this changed in 2019 with Star passing them both.
• If not Aya then Yuuka is at the top of PoFV, followed by Eiki and then Komachi
• Medicine is the lowest in PoFV but does relatively well overall
• Sanae is MoF’s best, often followed by Momiji
• Meanwhile Kanako is among the lowest. on that note, Suwako > Kanako
• Tenshi has the highest numbers of the spinoff characters
• Satori got way more yearly hits than Koishi until 2011, and since 2018 Koishi consistently has around half of SA’s hits per year
• Nazrin was very popular for a long while it seems
• Futo gets the most hits from TD, closely followed by Miko. meanwhile Mamizou remains at the very bottom surprisingly
• Of the Grassroots youkai, Kagerou quickly took the highest place followed by Banki and then Hime
• LoLK is a battle between Junko and Piece
• Sagume also gets a lot of attention huh
• Doremy had the most out of LoLK in 2018 (AoCF influence?)
• Sumireko had a small jump in 2018 too (VD?)
• Aunn and Okina quickly became the (hidden) stars of HSiFS
• Narumi had a huge jump in 2019. went right back down immediately after though
• Shion > Joon. though overall both drop off after their first year
• Mayumi had high numbers first two years, but dropped in 2021
• Yachie is on the rise
• About UM, Tsukasa got the most hits until the end of 2021, followed by Megumu
• On the lower end there’s Sannyo and Misumaru
• Yuuma is doing well

There's also some comparisons with the popularity polls but there's little correlation. And while I didn't count PC98, going by the numbers it should be about 15-18% of the Touhou tag (along with everything else I didn't search)

Tenkko, thank you very much! I can't imagine how much time and dedication this took to work out all the data. I appreciate everything and you deserve all the respect!

Those are interesting observations and I want to give some too throughout the next few days.

[CountVonNumenor]

Alright, so watching through all the research done for this thread, I am quite impressed about the amount of detail. I have one question though: have you only counted the safe-for-work art, or everything with the tag representing that character? Because I think you need to change some settings if you want to see the ABSOLUTE ALL bits of art from an artist (in my case, since I have the "hide anything R18" option active, I think I am missing in average like 20-30%, sometimes even 50% or more out of what a creator does).

Also, I think there should be counted the idea that there might be posts with multiple images in that post depicting the same character. Like saying there is a post with the tag "Reimu", but the post itself has multiple images of Reimu in it. Or group images, where you have too many characters to tag them all (max limit tag is 10 on Pixiv, compared to 30 on DeviantArt for example)



Like in this image. I have multiple 2hus, yet at how many there are, I did not manage to tag any of them. Therefore, the image only bears the "Touhou" tag in the upload


But for a bit of personal note, this is what I can come up with. Going by my favorite characters, the situation is a bit interesting:

• Hieda no Akyuu: 489 (a bit disappointed, but at least I brought my own contribution with an art piece there)
• Eiki Shiki, Yamaxanadu: 960 (also a bit low, but I can see why; again, contributed with one drawing)
• Renko Usami: 1486
• Maribel Hearn: 1252
• Youmu Konpaku: 5307
• Yuuka Kazami: 1947
• Mayumi Joutouguu: 895
• Raiko Horikawa: 394 (come on, I really think she deserves more)
• Tenkyuu Chimata: 1199 (maybe I can make it 1200 if I actually upload one drawing I have done a while ago)
• Utsuho Reiuji (Okuu): 1418

 

• (Bonus - former top 10 members, still enjoyed) Junko: 959
• Hecatia Lapislazuli: 831
• Iku Nagae: 679

[Isaac]

1 hour ago, CountVonNumenor said:

Alright, so watching through all the research done for this thread, I am quite impressed about the amount of detail. I have one question though: have you only counted the safe-for-work art, or everything with the tag representing that character? Because I think you need to change some settings if you want to see the ABSOLUTE ALL bits of art from an artist (in my case, since I have the "hide anything R18" option active, I think I am missing in average like 20-30%, sometimes even 50% or more out of what a creator does).

Also, I think there should be counted the idea that there might be posts with multiple images in that post depicting the same character. Like saying there is a post with the tag "Reimu", but the post itself has multiple images of Reimu in it. Or group images, where you have too many characters to tag them all (max limit tag is 10 on Pixiv, compared to 30 on DeviantArt for example)



Like in this image. I have multiple 2hus, yet at how many there are, I did not manage to tag any of them. Therefore, the image only bears the "Touhou" tag in the upload


But for a bit of personal note, this is what I can come up with. Going by my favorite characters, the situation is a bit interesting:

• Hieda no Akyuu: 489 (a bit disappointed, but at least I brought my own contribution with an art piece there)
• Eiki Shiki, Yamaxanadu: 960 (also a bit low, but I can see why; again, contributed with one drawing)
• Renko Usami: 1486
• Maribel Hearn: 1252
• Youmu Konpaku: 5307
• Yuuka Kazami: 1947
• Mayumi Joutouguu: 895
• Raiko Horikawa: 394 (come on, I really think she deserves more)
• Tenkyuu Chimata: 1199 (maybe I can make it 1200 if I actually upload one drawing I have done a while ago)
• Utsuho Reiuji (Okuu): 1418

 

• (Bonus - former top 10 members, still enjoyed) Junko: 959
• Hecatia Lapislazuli: 831
• Iku Nagae: 679

Like you said, it can tricky to work out the tags on Pixiv. There may be too characters on one artwork or there are multiple images for a single tag. I guess we can only estimate for how much more there is. I think it would be around 5% or less if there are too many in one.

[Tenkko]

2 hours ago, CountVonNumenor said:

Alright, so watching through all the research done for this thread, I am quite impressed about the amount of detail. I have one question though: have you only counted the safe-for-work art, or everything with the tag representing that character? Because I think you need to change some settings if you want to see the ABSOLUTE ALL bits of art from an artist (in my case, since I have the "hide anything R18" option active, I think I am missing in average like 20-30%, sometimes even 50% or more out of what a creator does).

I counted everything everything, sfw and r18. Actually thought about leaving out the latter since you could argue it's not "proper representation" of a character but eh whatever, it's a relatively small percentage anyways.

2 hours ago, CountVonNumenor said:

Also, I think there should be counted the idea that there might be posts with multiple images in that post depicting the same character. Like saying there is a post with the tag "Reimu", but the post itself has multiple images of Reimu in it. Or group images, where you have too many characters to tag them all (max limit tag is 10 on Pixiv, compared to 30 on DeviantArt for example)

I mean yeah this is counting tag hits. A character could also be barely visible in a pic and still be tagged. Some may not even be tagged at all and that's why this is approximate. One thing I did is attempt to remove the overlap within the groups, so those numbers refer to uh "amount of posts with at least one character from X tagged".

I should've mentioned that most of the time I tried to search the character's first name, since that usually nets more results than their full name, using the "Touhou" tag as a filter for anything from outside. One thing that happened that delayed the whole thing is uh, I first searched "Orin" and "Okuu", which get a few more hits the first few years but later on their full names are favored, so I had to re-do all those searches with the new tags for the sake of consistency.

ALSO my overall time range was 2007-09-10 - 2022-01-01. So the current numbers should be a bit higher.

[CountVonNumenor]

Nonetheless, good job with the research.

[Isaac]

1 hour ago, Tenkko said:

One thing I did is attempt to remove the overlap within the groups, so those numbers refer to uh "amount of posts with at least one character from X tagged".

I should've mentioned that most of the time I tried to search the character's first name, since that usually nets more results than their full name, using the "Touhou" tag as a filter for anything from outside. One thing that happened that delayed the whole thing is uh, I first searched "Orin" and "Okuu", which get a few more hits the first few years but later on their full names are favored, so I had to re-do all those searches with the new tags for the sake of consistency.

ALSO my overall time range was 2007-09-10 - 2022-01-01. So the current numbers should be a bit higher.

Great to know that you did all this to sure that the results were accurate! Looking forward to the other observations!