I see. Thanks! I can use this.

Since its async , I think what I’ll do is use the code to make a wrapper function for the dynamic import plugin, something like :

myDynamicImportWrapper( “blahblah” , default = “template-fusion-generator-dataset” )

Agree.

IMO The owner of a generator should be entitled to remove content from their image gallery

IDGAF about peoples NSFW preferences , but I’ve had occasional past experiences with users on my generator (fusion gen) who post content for the sake of harassment.

For example scat , geriatrics , bugs and gore , or just writing messages in the prompt. This happens rarely, thankfully.

But having the tools at hand to remove/hide such content would be good.

Thanks :) !

1. Download the vocab.json and find some way to sample it : https://huggingface.co/runwayml/stable-diffusion-v1-5/raw/main/tokenizer/vocab.json

Note that the vocab.json is badly configured for emoji symbols. They read as a garbled mess like ĮãĤĬãģŁãģĦ</w>

Ignore those.

Instead use emojis in prompt. Find some ways to sample these as well : https://perchance.org/fusion-t2i-emojis-2

All unicode symbols (not required , but can be fun to see what weird stuff you can create with these symbols) : https://perchance.org/fusion-t2i-unicode-2

You can try entering some words/emojis in this SD tokenizer , and you will see that they match the vocab.json list : https://sd-tokenizer.rocker.boo/

2. Divide vocab.json list two categories:

Suffix tokens = tokens that end with </w> , which represents whitespace. Like "banana " .

Prefix tokens = tokens without whitespace. Like “post” used in words like "postapocaplypse " or "postman "

Prefix tokens give properties to suffix tokens when placed in front of them.

3. If you have a means to invoke random prefix- , suffix- and emoji- tokens you can now mix them to create crazy results

I recommend using Notepad++ to sort the tokens.

4. Follow the cross attention rule: “SD reads the prompt left to right , one token at a time, finding association from the previous token to the current token (and the inage rendered thus far)”

You can try the -prefix . -suffix and the emoji-tokens out on the fusion generator if you like (I’m the dev) : https://perchance.org/fusion-ai-image-generator

There are also other datasets there which you may find useful.

There is a tutorial at the top, but its kinda wordy and haphazard.

I recently wrote a guide on Fusion Generator Discord for NSFW stuff : https://discord.gg/j6qAX3MWub

You can also find savefiles posted there with example prompts which can be loaded into the fusion-generator

I don’t want to present it as a “how to prompt” guide since there is no correct way to prompt things.

It’s more a way to show users that there are more options available to them , should they wish to use them.

Those issues are entirely subjective.

Are you even part of the furry community?

This sounds a lot like the rhetoric of someone who is an outsider that seeks to stir the pot because a group offends them.

As long as they don’t invade your space , don’t make an effort to invade theirs.

There are tons of generators on Perchance , each with their own crowd of people.

There is no reason for you to use the furry-generator , unless that is what you are actually interested in.

Please write about it here

What do you mean by “ill behaviour” exactly?

Are we talking people being abusive in the comments section?

Or is this more related to what people generate and post to the galleries?

I will do that. Big thanks for taking the time :) !

I posted some examples. You can try my generator if you like: https://perchance.org/fusion-ai-image-generator

I have savefiles for character portraits you can try , which can be found on this discord channel : https://discord.gg/6MvKGWAk

isometric level roleplaying

isometric knight character  reference sheet 

birdseye view of a knight  from above 

knight icon for tabletop game character 

8bit knight sprite turnbased rpg 

Feel free to browse the SD 1.5 token list for something useful: https://huggingface.co/runwayml/stable-diffusion-v1-5/raw/main/tokenizer/vocab.json

For example “dungeonsanddragons</w>”

and “tabletop” and “tabletop</w>” exist as tokens

Cool!

(And if @[email protected] is reading this , being able to add the .json save-file to the images in the image gallery would be helpful. Either within the EXIF metadata of the saved image png , or within a hidden “text field” that can store additional info apart from the text prompt, or something similiar)

Such a chat-feature would be really useful to me!

I just added a feature where users can save the generator prompt settings as a json file: https://perchance.org/fusion-ai-image-generator

So maybe copy-pasting the .json file could create a compact link or something.

Then another user clicks the link and it will load into the generator. IDK

Haven’t added this new chat format yet. Open to ideas.

Will try to give tips on usage to Beautiful People via Discord DM:s instead of here.

Discord link to 🔸✨📝🔰🔹 Fusion AI Image generator page : https://discord.gg/hSvEe5QK

Perchance discord server : https://discord.gg/A3SWynpT

For others who wish to use the token list in their generators , feel free to use the links below:

For testing:

You can test out the tokens in the 🔸✨📝🔰🔹 Fusion AI Image generator https://perchance.org/fusion-ai-image-generator.

For example , by writing " #commonSuffix# " in the generator you get a token from the “Common Suffix” subset of tokens.

You can write " #commonSuffix1# #commonSuffix2# .... #commonSuffix9# " to get up to 9 different tokens from the same dataset within a given prompt input field.

Optionally , you can sample the dataset using the 🔀 button in the generator.

Source for tokens : https://huggingface.co/openai/clip-vit-base-patch32/blob/main/vocab.json

Generators:

https://perchance.org/fusion-t2i-sd15-clip-tokens

All 47K tokens split into prefix/suffix groups:

https://perchance.org/fusion-t2i-sd15-clip-tokens-prefix

https://perchance.org/fusion-t2i-sd15-clip-tokens-suffix

All 47K tokens split into prefix/suffix and rarity in the training data (size of their ID number , where exotic == highest IDs):

https://perchance.org/fusion-t2i-sd15-clip-tokens-common-prefix

https://perchance.org/fusion-t2i-sd15-clip-tokens-average-prefix

https://perchance.org/fusion-t2i-sd15-clip-tokens-rare-prefix

https://perchance.org/fusion-t2i-sd15-clip-tokens-weird-prefix

https://perchance.org/fusion-t2i-sd15-clip-tokens-exotic-prefix

//----//

https://perchance.org/fusion-t2i-sd15-clip-tokens-common-suffix

https://perchance.org/fusion-t2i-sd15-clip-tokens-average-suffix

https://perchance.org/fusion-t2i-sd15-clip-tokens-rare-suffix

https://perchance.org/fusion-t2i-sd15-clip-tokens-weird-suffix

https://perchance.org/fusion-t2i-sd15-clip-tokens-exotic-suffix

Question ; If I add this feature to my AI image generator https://perchance.org/fusion-ai-image-generator, and people choose to use NSFW backgrounds in chat , is that going to cause problems for site ad-monetization?

Also ; can we add a “comments” button to images posted in the AI image gallery?

Having access to that list could be useful 👀

Great suggestion! While we are at it , I’d like some $meta options for the “perchance screenshot” of the generator.

For me specifically , either;

1. to include a $meta image link to use as the “perchance screenshot” , or

2. to include a $meta setting for how far down to scroll before taking the “screenshot” of the generator

I learn new stuff about Lemmy everyday. Thanks mate :)!

Update from last comment: Things from my NSFW image gallery where the #studio# and #name# datasets are used , cuz I want to contribute.

( Links to datasets are in my other comment, should you want to use them for beautiful-people , allo )

Nice!

Let me give you some good stuff for your generator , allo.

Studios dataset : https://perchance.org/fusion-t2i-nsfw-studios-dev

Pornstar Name dataset : https://perchance.org/fusion-t2i-nsfw-names-4

Start the prompt with an item from the #studio# dataset and include the nane somewhere , e.g

#studio# … #name# … with xvidsnudity

(or adultnudity , or something else) .

There are lots of prefixes you can put in front of tokens like “nudity</w>” etc.

Haven’t made a dataset for these prefixes yet.

I can’t post explicit stuff on my generator since I’m the dev , but I have made some nude stuff on my NSFW gallery to showcase this

The penis problem is very real!

My E621 dataset can do dicks really well , but the photorealistic category is struggling lol :)

Link to part of the E621 dataset , if you want to use it for your generator , allo : https://perchance.org/fusion-t2i-e621-tags-6

I’m thinking of making an all-male sub-category for photorealistic and anime NSFW at some point