Wasn’t the gargoyle behind a leaf supposed to be inside a box?
That’s so beautiful! Thank you!
Something like that yeah. Personally I could never get from ‘picture’ to gargoyle/が but have no trouble getting to ‘picture’ from 画 so the box didn’t really matter to me.
I just got that kanji out of apprentice too so it seems to be working!
I really love this thread! I’ve kind of just been a lurker this whole time but I still am totally amazed by every new image the AI creates. Thanks for creating it!
Thanks to the userscript @Sinyaven has already written it was easy to just connect the script with a Github page where all the images can be uploaded that you want to have as mnemonic images. That way the userscript does not have to be updated every time someone wants to add new images! 
I have already uploaded most of the images currently in this thread on this Github page I made. If you want to add images just send a pull request (I don’t know if you can make Github repositories publicly editable without needing pull requests?) or ask me and I’ll add them to the folders.
You can also download the script here (you will need to have the WaniKani Open Framework installed). It is 99% written by Sinyaven I just tweaked it a tiny bit.
Download:
Here is an example of how it looks on the info page for 足りる:
Or for 魔:
Here is an image for the review of the radical 矢:
Wow, that’s really cool! Thanks for doing this
Thank you for lurking 
I’ll definitely try to use the script. You know what, I’ll try to use it right now 
brb
It’s a little embarassing but I feel like I’m too stupid to use the script aka I never worked with github and don’t know anything about userscripts. I’m as far as having installed the script and now whenever I’m on a page where I can add a hint, I see that tampermonkey is active. But now I don’t know what to do. I also searched for an explanation and wasn’t able to find one. Could you tell me where to go from here? 
So sorry to be as annoying but the script is such a cool idea and I’d love to use it.
Oh, yeah no problem!
You can’t add the images directly on the WaniKani page. But everytime you post new ones here I will add them to the database (which is on Github but you don’t have to worry about that, I can do all the uploading and stuff).
So, on info pages for which there already is an image in the database it should show it under the reading section. But you have to have another userscript installed for it to work. That one is called WaniKani Open Framework and it can be installed like any other userscript here: Installing Wanikani Open Framework
If they are both active you should be able to see the images (for example on this page: WaniKani / Kanji / 画).
Of course, there aren’t that many images yet but as we all make more it hopefully also gets more useful as time goes on!
Hope that helped!
You are A HERO <3 Thank you.
I hadn’t installed the Wanikani Open Framework - literally the first thing I should have done. lol
Now it works. 
Glad to hear it, hope you enjoy it! And thanks again for all the great images! 
I love the idea of using AI generated images from mnemonics. So great.
This has the earmarks of another “breakout” application for AI (like natural language processing). I expect we will see a lot more of this stuff going forward.
I just watched an interesting video on the topic here:
More eye candy re: Parti specifically here: https://parti.research.google/
Cool! Thanks so much for sharing 
So I’m in charge of creating images if someone used up their credits on midjourney / has no account and you’ll upload the images to the database. #dreamteam
And again, to anyone who would like to get their kanji, radical created, just comment and I’ll take care of it asap
I finally ran out, lol. I think I’ll eventually buy a subscription though. One I’d really like to see is: 災 (disaster) - river of fire. Should be interesting.
From my wall of shame: 免れる (to escape something).
The (slightly reworded) reading mnemonic is “you jump inside your manufactured car which is a car for a tough man that needs to drive fast, just like you”.
I’d be very curious to see what an AI makes of that.
Haha, true!
Wow, that’s really interesting. Not only images but also human-like text. That’s really impressive 
That looks cool as well. Could’ve been created by this AI judging by style!