Book of Monsters
About
A collection of 2000 AI generated monsters and their descriptions. Each entry has an image with a description of the monster. The AI was trained on images from D&D. This is mostly just for fun, but maybe it can give you inspiration for a story, game, campaign, etc. Feel free to use anything as long as it is non-commercially. You can download a pdf for offline viewing, be aware that it is somewhat big (360mb).
For a little easter egg, press S on your keyboard to get an extended description.
Disclaimer
Seeing as this is AI generated there is always a risk it generates something weird. I have gone through a bunch of it and automatically filtered out a lot, but there is always a chance I missed something. So read at your own discretion, and if you find something you think doesn't belong, please let me know.
Tech & Credits
- Images generated by training a ruDALL-E model on D&D images, with the help of Looking Glass.
- Image tag identification using Google Vision AI.
- Name generation by Fantasy Name Generators.
- Text generated using GPT-J.
- UI borders by ScrewyLightBulb.
I might add more monsters in the future if people like this, but for now 2000 should keep you busy for a while :)
Thanks for checking this out! If you liked it, a comment or rating is always appreciated!
Download
Click download now to get access to the following files:
Development log
- PDF versionMay 10, 2022
Comments
Log in with itch.io to leave a comment.
SCP2.0 :skull:
462 has to be my favorite.
Probably the strongest being in all of media.
Lol, beware the Cholanga.
Hi!. Awesome work!!.
Is there any option or form to insert that into pixel format with any software or technique?
Thanks again.
Thank you!! There is no option for that, but pixelizing should be pretty easy with image editing tools, or some online tool. (for example, this seems to work pretty well: https://www.pixelicious.xyz/ )
I would love to learn more about the development process, as I'm really interested in making something similar.
Yeah of course, any specific questions you have?
For now I can say that the basic workflow was:
1. Collect dataset and train an image AI (ru-DallE). Generate Images.
2. Use google lens to find names of interesting things in the image to possibly use in the description.
3. Use GPT-J on a cloud TPU to generate the descriptions with results from step 2.
4. Final filtering of images/text, then make this app. The images and descriptions are hosted for free on cloudflare.
I'm guessing #2 is the implementation of Looking Glass? Or is that used elsewhere?
And I suppose, for the most part, the workflow the main thing I wanted to know about. How much did you train the AI prior to building the app? And was there a reason to use ruDallE over just DallE?
As well, did you have much experience prior to this with AI? With how quickly AI had developed in the last few years, it's been a bit overwhelming to try and find resources to understand everything.
Yep, I used looking glass for that.
I chose ruDallE because at the time at least it was the only available version of DallE that was open source and I could train myself. Not sure if there are other versions of DallE right now that allow custom fine-tuning. If I remember correctly I did 20,000 iterations, not sure how many epochs that was. I think in total the training time was around 8-9 hours.
Basically no experience with AI before this, it's indeed pretty difficult to start out, and I think right now I could already do a lot better with DallE2 and ChatGPT existing. Unfortunately I can't really point you to good resources to learn, I just looked for different info all over the place.
Hope this answers some questions though!
I know you mentioned about can't point to good resources, 3 months later. Is there anything you can share to get started doing something like. SDXL is out and I believe on civit ai there are a lot of LoRAs thatn ca be used to train stuff like yours. I wanted to know what did you use as far as hardware GPU to train it.
I don't really have any experience with the more modern tools like stable diffusion, so I still can't really give good advice on that. My personal GPU is pretty old so I used sagemaker studio lab which gave me a pretty good free GPU to use (I think it was a tesla v100). Google Colab (pro) should be a very good option right now as well.
73.............
Nothing to see there, just a harmless plant.
yes sure)
Is page 1329 "Kangephant" intended to have this description?
"Kangephant is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it. It is a being that brings happiness to all that see it."
I don't know if it's supposed to be repetitive like that or not just thought I'd point it out
Lol, well all the description were written by an AI, and since the AI I had access to wasn't that good I guess some descriptions are like that. I guess it makes the monster a little bit more creepy that way haha.
Kangephant has some incredible abilities. For example it can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously. It can see every single person in the whole world simultaneously.
#679's picture is kinda scary lol
Yep, that's an interesting one.
#1777 is a sneezy death bunny
Haha. Kinda unfortunate that it has to sneeze to locate it's prey.
Thanks for the suggestions + numbers, gonna check them out.
Glad you like them! Yeah some game with it would be awesome to see. And the cool thing about AI is that it can just keep generating near infinitely.
Achuksuh
page 611
If he is sent some food by a person, he will say "I'm full, but I'd love some more.
This creature might seem like an ordinary flowering plant, but is actually the cutest little thing you've ever seen.
He lives in a small garden made by a man named Sna-Gurth. Achuksuh is friends with another furry creature named Sna-Methos. He is also friends with the "honeybee" creature named Sna-Ree
Achuksuh and Sna-Methos are good friends and like to go out and find flowers and other things to eat. However, their main hobby is to go out and look for other creatures.
Achuksuh is a being that brings happiness to others. He is the kindest of creatures and likes to eat flowers and berries. He will befriend any creature that he sees, and he is also very friendly and kind to humans.
amazing. thank you for this. Achuksuh brought me happiness.
Wow, that's probably the cutest description in there, also love those names.
And I guess if Achuksuh brought you happiness that means the description is true.
Is it possible to put a download to this I would love this (I have already spent about an hour on this)
Haha, nice to hear you can get lost in it. I added a pdf to the downloads that contains every monster. Might be somewhat slow to load but hope it works.
Thank you I will have fun looking through them
This is so cool! Can't wait to spend hours lost in this. <3
Let me know if you find something cool!
As always, you surprised me! Congratulations :)
Haha, great!
fantastic！
thanks!
This is so great. Any possibility of a downloadable PDF version?
Oh, that's an interesting suggestion! Might be possible, gonna have to look into it.
Glad you like it!
Thanks. The interface is a bit wonky in different browsers, and I'd love to be able to zoom in on the art more easily. It appears the consistent visual framing of creature against plain white background gave the GAN a lot to work with effectively.
Ah, I tried to get the UI right for most devices. For what browser is it wonky for you? Zooming function is also a good idea.
Yeah I'm really surprised how well the GAN actually performed, especially with a somewhat small dataset.