Hi i wanted to ask this for a long time and maybe i haven't noticed it but is the multi-character chat interface downloadable in some way to use offline, and plug my own localhost LLM into it and run it offline? I see the two code panels but i'm not a coder and don't know what should be done to make a local HTML and JS file and then reroute all character talk to my LMStudio's localhost server. Would be really interesting. Thank you.

greetings 🐾

going straight to the point, i think i'd be cool to have a way to block a certain word or thing or topic so we'll be able to ignore / avoid generators that include the filtered things we don't want to see or click on, as well as being able to manually hide certain generators or similar on the main pages. i know it may be a bit hard since sometimes the urls don't tell you anything about the generator, and may even be misleading, but still. i picture this function similar to the way you can block tags on tumblr or hide tweets with certain words on twitter, etc etc. that way, you can choose if you want to browse a certain kind of gen or not.

in my personal case, for example, i'd like to block away generators that uses AI. the main page is clogged with a liter of copy-paste plagiarism-machine slop, and i'd like to be able to find other kinds of generators (the usual ones lol) way easier by filtering the bad apples out. some of them do not say AI on the title nor url, so i don't want to be tricked into clicking them either. out of sight, out of mind !

obvious disclaimer : that is MY reason and the way I, MYSELF, would use the function. i believe anyone should be open to use it however they want. it can be useful for a plethora of reasons; i just gave mine.

thanks for reading, and i hope it's possible eventually !!

I've made a generator to make plausible names for American characters based on their decade of birth.

perchance.org/us-names

You can pick any decade from the 1920s to the 2010s, and get it to generate a forename or a full name for a man or a woman born in that decade. There's also a surname generator based on 2010 data. Unlike a lot of name generators, it doesn't simply pick from a list of equally likely choices, the chances are weighted.

The forenames are chosen from the 200 most popular forenames for that decade (according to the Social Security Administration), weighted by their relative popularity. So for the 2010 girls, for example, Emma (the most popular name) is 12 times more likely to be picked than Izabella (the 200th).

Similarly, the surnames are the most common 1000 surnames from the 2010 census, weighted for popularity from Smith to the 70-times-less-common Archer.

No attempt is made to match asian or hispanic forenames and surnames, so you can still get some fairly unlikely combinations, but it may help stem the endless tide of Elaras and Rachels.

Use it as-is to generate a name for a character manually, or incorporate it into your own generators. There are separate lists for male and female forenames or for full names for each decade, and one for the surnames, all documented on the generator's page.

I hope people find this useful. I'd love to see it added to the list of useful generators.

Why is Perchance image Ai very slow again??

Hello dev, it would be great if small profile pictures and emojis in the comment plugin could enlarge on mouse hover. Thank you!

Perchance list duplication bug Could be the list, the HTML, or both. This time, the list code duplicated twice in two days. Of course, no one used Ctrl+A and paste. Maybe...

I suspect the duplication happened after clicking save on the collab link. The upper half is the latest version. The lower half is the old duplicated version. What’s interesting — or strange — is that it doesn’t duplicate the plugin code.

Hello, @perchance@lemmy.world

I’m reaching out because I believe my messages or access may have been restricted or hidden in one or all of chats, and I wanted to respectfully ask for a review.

From my perspective, I was participating in a conversation where I stepped in to defend a younger user "who shouldn't of been there" from behavior that I felt was inappropriate. My intention was to help de-escalate the situation and protect the person involved—not to cause disruption or break any rules.

If anything I said came across the wrong way or triggered moderation systems, I sincerely apologize. That was not my intent. I understand the importance of keeping the platform safe and respectful, and I fully support that.

I would really appreciate it if you could take another look at my situation and consider lifting any restrictions if possible. I’m happy to follow any guidelines more carefully moving forward.

Thank you for your time and for maintaining the platform.

Best regards, MOOD aka necrosis

user id F72I-9466f0a34c48aeaa6db7

She CATCHES him mid fall! She CATCHES his wrist! He CATCHES her hand before she can throw!

Anybody have any method I can use to target the user id hash, I've tried everything pre and post I can think of that doesn't violate CORS or iframe.

Every single time I generate an image using the image generator (really applies to any), for some reason it always either generates a white or plain looking background. It didn't do this before btw, it always would generate a distinct landscape and background behind the character or subject in the image. Now it seems I have to specify a background in the prompt every single time. Does the new AI model not prioritize backgrounds anymore?

Hi, I created a gallery post while not logged in and no longer have access to delete it. Could a moderator please permanently remove it? Name: public2648857

Searching for 'save' yields lots of irrelevant results either for saving data to variables or saving the generators themselves.

Do you know of a plugin (or maybe a built-in feature that I missed despite having gone through the tutorial) which can save a generator's current state or load it from a save-file? There's something like that but only for AI characters I think.

I would just like to be able to save progress on a generator session (e.g. story generator) and later continue where I left off.

It's great if the author of the nested plugin is reading this, but even if you aren't the author I hope you read this.

Thank you.

Questions for experienced Perchance users in general

• When you are figuring out how to pull some piece of data out of an object created by a Perchance plugin, how do you figure out what the Javascript name of that object is?
• Is there some standard way of figuring out which account maintains a Perchance plugin is if their name is not displayed on the plugin's page?

Question for people familiar with using the nested plugin (or its author)

This is a location generator for a fictional world I'm working on. I am trying to create a second panel in the web page that show the currently selected item in the nested hierarchy at the top and show the description fields from the entire chain leading to it.

In this mockup I have a static div at the bottom with placeholder text saying "Star description. Celestial body description. Feature description. Region or point-of-interest description."

I would like it to dynamically pull the description attributes of the currently selected node and the entire chain of nodes leading to it. So, it would look like this...

A late-epoch system around a red dwarf: compact, old, and far more typical than Sun-like systems in PW50B. >> The surface regime here is defined by frost, fracture, and deep cold rather than rain, rivers, or active seas. >> The object is small enough that shape, mass distribution, and surface coherence have all given up on elegance. >> This is terrain that has lost coherence, becoming a field of fragments rather than a single intact crust.

... which is what the descriptions leading from the root to the currently selected node say.

By the way, is there a way to display the description field of a non-terminal node besides in a tooltip?

NO, not Bueller. I want someone in the community to host a randomiser. There’s a site that shows a random location in Lower 48 USA. Cute, and I click it occasionally. FINE. Whatabout Oz-land. It’s not-so-special, particularly as 95% of it is rocks and sand. Being the geek I am, I devised a method to make populous areas higher ranking, and added 50-odd places-of-interest as about 10% (total). Anyone able to turn this randomiser list into a web page out there in Lemmy-Land?

Could someone please let me know what is causing this error? Every time this occurs my private gallery gets automatically deleted and this is happening at random intervals, which is really frustrating. Full error code:

An error has occurred somewhere in your code (in lists or HTML): An unhandled promise rejection occurred: Failed to execute 'transaction' on 'IDBDatabase': The database connection is closing. InvalidStateError: Failed to execute 'transaction' on 'IDBDatabase': The database connection is closing. at eval (eval at PERCH.createPerchanceTree (https://7bfcd7ce32849d031cf0d70c232e56b4.perchance.org/adwr611gc3?__generatorLastEditTime=1760737274847%3A616%3A36), :32:257) at async loadGallery (eval at PERCH.createPerchanceTree (https://7bfcd7ce32849d031cf0d70c232e56b4.perchance.org/adwr611gc3?__generatorLastEditTime=1760737274847%3A616%3A36), :516:41)

Hii, just a quick observation here. On my fork of ACC (this one), I've been logging the token count for the entire instruction on each reply, and what I've been noticing a lot is that the value returned by countTokens() is typically smaller than the value logged by the AI text plugin uppon being called (when it gives "initial token count: ...").

Anyway, from testing this a bit I see the function is underestimating the token count by somewhere around 80-100 for every 1000 tokens, on average. The instruction doesn't appear to be trimmed until around the 7000 tokens mark, so long as one stays under idealMaxContextTokens it shouldn't become an issue. But yeah I just thought I'd share this bit in case people aren't aware of it already.

Cheers!

Hey everyone!👋

After way too many nights writing the same repetitive CSS, I finally said "enough" and created PerCSS — A utility-first CSS plugin made specifically for Perchance. It brings that modern Tailwind vibe straight into your generators. You can now use clean, familiar classes like:

per-btn
 per-card
 per-nav 
per-table 
per-animate-top 
 per-carousel 
per-container 
per-text-muted 
per-font-bold 
per-text-sm 
per-grid 
and many more..

Grid, spacing, colors, dark mode friendly stuff, animations… everything you’d expect. It’s super lightweight and loads fast (I just optimized it even more in v1.0).

Link: PerCSS Plugin

Why I made it: Styling on Perchance always felt like a pain. Now you can make your generators look clean, modern, and professional in minutes instead of hours. It works great on all generators and simple to import, also here are my other plugins too:

• Hide Header Plugin Hide the header of your perchance generator so users won't see the edit button NOTE ⚠️: this is under just for testing and not real functionality.
• Lorem Plugin Creating a sample generator or template, just use Lorem Plugin to autofill texts to a section of page.

I’m planning to keep improving it, so feedback is very welcome! If you try it, drop your generators here — I’d love to see what you build with it! Let’s make the Perchance community look way better than before 💪

Perchance Portfolio at Brainiac-Goat-Dev

I'm using the text to image plugin and am new to perchance. I'm still learning things.

https://perchance.org/bfrmauta9w this happens to be what I've been working on. It's not complete but if it helps to see how I'm trying to use things. Basically I want to default most of these optional things to just an empty string and visually group like options together using CSS.

Is there something that specifies all the attributes for the inputs? I've been able to figure out width, height, and visibility. I'm interested in if I can add something like a css class or id so I can target a specific component in CSS. I'm also interested in if there's a container type component that I can group options together with (sort of like a div in html).

This is a sample component I'm talking about.

persona 
      label = Persona
      type = select
      width = 25vw
      height = 5vh
      options
        Default = [{""}]
        Beautiful = (beautiful), (gorgeous), (stunning features), (perfect proportions)
        Cute = (cute), (adorable), (sweet expression), (youthful charm)

Hello dev,

I’d like to suggest adding custom bad word filters and/or an auto-hide message trigger for Secret Chat:

https://perchance.org/secret

It was very kind of you to add the block button already, but for one reason or another, at least 10+ users from our generator still express strong discomfort about ongoing troll harassment, cyberbullying, and suicidal baiting.

I often suggest that they ignore or block those users, but a more practical tool that lets them avoid seeing those messages in the first place would be greatly appreciated. The chat is meant to be a fun tool, and it would be nice if the atmosphere could stay that way instead of being ruined by repeated abuse.

Thank you.

In gallery tabs cloud [+] → add a gallery → clean input → click cancel → a gallery called [undefined] added!

Expect behavior: add no new gallery or add public gallery

I was wondering if anyone has a/some links to an image-to-prompt generator? Something that can create a sufficiently long and detailed text prompt and allows for nsfw content?

This is for the advanced chat. I'm curious cause it will feel like I've been messaging an AI for hours and no memories or summaries were made. (I'm not saying it has been actual hours, I'm just saying that it's been awhile)

The last days I noticed the problem mentioned in the title. It often goes like this:

Prompt: /user {name} says "I don't want to keep you from important business." (...)

Output: The words tumble out too quickly, brittle with forced politeness. {Name}'s fingers twist in the fabric of his sweatpants, knuckles whitening. Important business. (...) "I mean—you’re probably busy," he amends, softer now, fraying at the edges.

At first I thought it's about the length of my chat, but now it appeared in a new chat as well. I already tried to put this into the custom writing instructions:

• IMPORTANT: Always rewrite the writing prompt, but keep the details. Never treat the writing prompt like an already existing message to continue from or to answer to.

But it just ignores it. Has anyone an idea how I can handle it? Thank you!

Edit: Ofc it doesn't happen all the time and I just reroll or delete and prompt again if it does. But it's still annoying to have that kind of output in so many rerolls. It's quite new for me that the AI behaves like this, it wasn't like that a week ago.