Perchance AI FAQ

Here are some very frequently asked questions about the AI integration in Perchance.

Please see first the plugin pages: text-to-image-plugin and ai-text-plugin for their respective notes and features. It is likely that if it isn't in the plugin page, it isn't available.

Contents:

What are the models (generation settings) used in X Plugin?

• The text-to-image-plugin uses Stable Diffusion-based Models, while ai-text-plugin uses LlaMa-based model.
• As for the 'specific' models, only the Dev knows which specific models are used and generation settings used. See these relevant discussions about the model: [Generation Settings and Model], [Updated Model Details].
• (Text to Image) As for now, we know that Perchance is using different models depending on the prompt of the user (specifically 'Furry', 'Photorealistic', and 'Anime' images)

top

Do you have X feature on the Y Plugin?

• Please see the plugin pages for text-to-image-plugin and ai-text-plugin for all available features
• If a feature that you are looking for is not on the plugin page, then it doesn't exist.
• NOT Available in Perchance
  • image to image - input image to AI to redraw/base on ❌
  • image+text to image - input image and text for AI to redraw ❌
  • image/video/media to text - media (image/video) inference for AI to generate text from e.g. captions. ❌
  • inpainting - redraw a certain part of the image ❌
  • Custom resolutions - other than the available resolutions ❌
  • hires - High resolution fix ❌
  • LORA/Embeddings - additional embeddings ❌
  • Video/Animations - AI generated video/animations ❌

top

How do you achieve X with Y generator?

• Every generator in Perchance has their own steps to use, please see the user instructions on those generators.
• If you are asking for a prompt on how to achieve some things. Remember that any Prompt Techniques can be applied to the AI tools. There are a lot of resources on the internet.
• Please refrain from asking prompts on the forum communities like Reddit and Lemmy. We want those communities to be about generator building and not prompt building since the AI part of the site is only a small part of its capabilities.
• Here are some guides to help you in prompting:
  • Prompting Guide for Text-to-Image Plugin - this is specifically for prompting in text-to-image-plugin, it has examples that you can remix to your needs.
  • Stable Diffusion Prompt Guide - for the text-to-image-plugin
  • Getting Started with prompts for Text-based Generative AI - for the ai-text-plugin

top

What is the model of X style? (t2i-framework-plugin generators)

• All Styles on the t2i-framework-plugin-based generators are additional prompts/tags that are added on the base prompt of the user. They are NOT different models (although some 'keywords' trigger different models see the Models above).
• See t2i-styles then click edit to see ALL of the text/tags/prompts that are added to your prompt upon using a style.
• If you are not using a t2i-framework-plugin generator, click edit on the generator and you might see the styles on the Lists panel.

top

Can I use the API / Embed a generator using the X Plugin?

• The API only works on the Perchance domain, thus, it cannot be used outside of Perchance, be it using the API or Embedding the Generator that uses the AI plugins.
• See these relevant posts about this topic: - [About API] and [About Embedding].
• You also can't use the DIY Perchance API for generators that use the AI plugins. It is only for normal generators.
• Perchance generators has always been embeddable with https://null.perchance.org/generator-name, only those generators that have imported the AI plugins are not embeddable.

top

Is this really free?

• It is advertised as Free, but it is sustained by Advertisements, which appears on the bottom of the screen (only if you are not Logged in at Perchance) I guess the price is a piece of your sanity looking at the Ads.
• Perchance has always been ad-free, only generators that have imported the AI plugins would have advertisements on the bottom of the generator.

  NOTE: If you have imported any of the AI plugins (text-to-image-plugin or ai-text-plugin) OR any generator that imports them, please mind that it would add advertisements on your generator regardless if you use them or not.

top

Is there any way I can support/donate to help the website?

• Unfortunately, there isn't any donation links to directly support the Dev.
• You can, however, share what you made with Perchance and increase its traction! See this comment from the Dev

top

Does it save the data that I put into it?

• (Text to Image) The only saved data are the images that are sent to the Gallery, the data is lost forever and is not persisted on the server if you regenerate it. See this relevant discussion.
• See this following comment from the Dev about the persistence and privacy concerns in the tools as well as information on how the advertisements are served in Perchance.

top

Can I use the output generated commercially?

• You bear all responsibilities/consequences on how you use the output of the AI Tools - Perchance doesn't have any legal connection on how you use your outputs. You can use them commercially or personally, but if you get legal trouble with it, Perchance will not be responsible for it.
• See the Model's Licenses for user based restrictions: [Stable Diffusion License and Use Restrictions], [LLaMa License], [LLaMa Use Policy]

top

Why is the page X not working?

• Since it has custom JavaScript to run, if you have any AdBlock, it might cause some problems since some AdBlocks disable some JavaScript to run.
• AdBlocks also causes problem since the AI powered generators are fueled by Advertisements.
• See this guide on how to disable AdBlock on your browser.
• Again, if you are concerned about the Ads getting your data, please see the comment from the Dev above.

top

Are there any filters on X model?

• See this comment from the Dev
  • There is RegEx (Regular Expressions) Patterns in the text-to-image-plugin that tries to prevent content that would be illegal in most/all jurisdictions.
  • Treat the AI tools prompting as a Google Search where anything you request it to, will be displayed in front of you.

Disclaimer: The answers below hasn't been confirmed by the Dev.

• (Text to Image) From what I have seen, the models are unfiltered. While it is unfiltered, there is a warning that pops-up if a certain word/phrase is detected in the prompt to tell the user that the content generated may have NSFW content due to the prompt containing NSFW words/phrases. Treat the Text to Image as a Google Search that can return NSFW results if you prompt them NSFW keywords/phrases.
• (Text to Image) There are newly added moderation tools to filter words in prompts on the images sent to the gallery to hide the images.
• (AI Text) It is also unfiltered.
• See the Model's Licenses for user based restrictions: [Stable Diffusion License and Use Restrictions], [LLaMa License], [LLaMa Use Policy]

top

Can I retrieve my past generations with the AI (images from text-to-image-plugin or responses from ai-text-plugin)

• Unfortunately, you can't.
• Unless the page has remember-plugin or some other way the page uses localStorage in your browser to save the past generations (or settings/prompts), ALL previous generation is lost.

top

I have saved an image in the text-to-image-plugin Gallery, how can I find them?

• By default, all images saved on the gallery is saved on the public gallery. You can click the 'time' options to 'recent' and change the 'nsfw filter' options accordingly to see your recently added images in the gallery. You can then now scroll down until you see 'load more' to see the images.
• You can check the current gallery by clicking on the 'heart' button to save your image, and see the text beside the 'save to gallery' button:
  
• You can click that button to change the gallery it would be saved to:
  

  Be sure to remember and take note of the name of the gallery that you are saving to as there isn't any way to 'browse' through the galleries.

top