There seems to be a problem with ‘installing’ the required packages (something like stdlib) for it on the web, even if the script itself is loaded. @perchance pinging dev so they can take a look at it.

Can we do freestyle - no specific topic?

You could use ShareX screen record then upload to Streamable. Can you also provide the thread/chat export that is faulty so I can try to replicate the problem on my end?

I’d recommend waiting for a couple of seconds before reloading/closing the page to make sure the image is ‘converted/saved’ on the chat. I tried clicking the keep then immediately reloading the page (even closing the page as soon as I clicked ‘keep’), but it managed to keep the image. Not sure what other factors might affect it saving.

If you could provide a video demo of the problem with the console open, it would be helpful to see if there are problems with the code.

Might need to update the Perchance Data.

Glad to help!

After the declaration of ‘odds’ any other text after it is invalid e.g.

Ss^0.5 or s
Tt^0.5 || t

If you want a randomized value for that in which Tt has odds of 0.5 to being chosen you could have:

{Ss^0.5|s}

{Tt^0.5|t}

If you want to add ‘odds’ with HTML formatting, you should place the odds after the items e.g.

<u>Ag^0.5</u>

to

<u>Ag</u> ^0.5

The change in odds should always be on the end of the item.

On your [m = markings6.selectUnique(1,7).joinItems("/")] you are setting the value of m to be the joined items in a string, not the list. To fix it you just need to store it on the variable, then use the joinItems on the variable after: [m = markings6.selectUnique(1,7), m.joinItems("/")]

Then to output the names: [m.selectAll.map(a => a.name).joinItems("/")] in which it would iterate on the markings on m and get their name, then joinItems them.

🎃

Sararari Catteyon - Vionet20

You and your group of friends stumble upon Sararari’s lair, where she presents you with a grim ultimatum: eat her demonic offerings or watch each other starve. She observes with a chilling calm as you grapple with the horror of your predicament.
Character Chat Link

@perchance@lemmy.world

The link of your generator is the URL of your page: https://perchance.org/your-url-page

Thanks 🙏!

I haven’t really had a long conversation and multiple topics in one bot, so I don’t have much experience in manipulating the bot’s behavior.

I would suggest adding a reminder on the character’s Reminder Message to steer the AI’s response. Another would be using the /ai <reply instruction> method in which you instruct what the AI would reply. You could also probably try to edit the Memories and summaries that the AI generated that it uses in generating the reply and you could check the memory that pushes the AI to reply that way by clicking the brain icon that show up upon hover in the message on the bottom right.

There are a lot more people knowledgeable in telling the AI what to do in the #ai-character-chat channel in Perchance Discord.

Nebula and Forest

generator: text2image-generator
style: No Style
prompt: nebula in clear night sky framed by forest leaves in a clearing, detailed, soft lighting, rtx, uhd, anime style, featured in pixiv, WLOP, ross tran, artgerm
negativePrompt: \[3d render, vibrant, child, low contrast, cel shade BREAK hard shadow, unclear features, incorrect anatomy, deformed body, bad drawing, poor illustration, worse quality, cellphone:0.1\] 
resolution: 768x512
guidanceScale: 7
seed: 187151457

@perchance@lemmy.world - pinging dev for other info that I’ve probably missed.

Possibly missing ending quotations?

Before the "negative", you didn’t close the quotation then you used a dot instead of a comma in transitioning to the negative

...
"Style Name":{
  "prompt": "...",
  "negative": "..."
},
...

If you are trying to run the ‘ai-character-chat’ on glitch, you wouldn’t be able to do so. Since AI plugins doesn’t work in other domains, only in Perchance. I suggest looking into the OpenCharacters which is Perchance based off of.

I tested it, and it seemed to be working properly, possibly something on the AI’s response inside the <image> tag that borked the code?

AI Celebrations Quiz

https://perchance.org/ai-celebrations-quiz

An AI Powered Multiple Choice Question about Celebrations in the world.

Can you send me the character link again. I would assume that you do not have [input.description] on the prompt or [input.negative] on negative on the pasted style like so:

  ...
  "Professional Photo": {
    // When adding a new prompt, you need to add [input.description] to help the code know what would be the prefix and suffix of the prompt. [input.negative] is not as required.
    prompt:
      "[input.description], {sharp|soft} focus, depth of field, 8k photo, HDR, professional lighting, taken with Canon EOS R5, 75mm lens",
    negative:
      "[input.negative], worst quality, bad lighting, cropped, blurry, low-quality, deformed, text, poorly drawn, bad art, bad angle, boring, low-resolution, worst quality, bad composition, terrible lighting, bad anatomy",
  },
  ...