Welcome to my personal, quickly made, guide to prompting and using Illustrious. I will keep everything simplified as possible yet explain them as much as I can. Take everything here with a grain of salt, as I may have forgot some things, or I said some things wrong. If I forgot anything or said some things wrong let me know what they are and how to fix them as I'd like to improve this guide. Before we begin the clip of a checkpoint model is the main knowledge it knows.

Section 1)

Section A) Basic Knowledge

Illustrious is a fine-tuned SDXL model, like NoobAI and Pony
Illustrious is trained on Danbooru Tags, unlike NoobAI which is trained on both Danbooru and E621.

If the model has Illustrious 2.0 or Illustrious 2.0 Stable in it you can run a tad bit of NL (Natural Language) aka Captioning like you'd do with flux, just only a little bit as it might not understand the full thing. If you are using a model with versions of Illustrious before Illustrious 2.0 like WAI-illustrious-SDXL it won't know the same thing as Illustrious 2.0 and Illustrious 2.0 Stable. Hyphoria however for example is Illustrious 2.0 due to it's Clip (Knowledge) being Illustrious 2.0, meaning you can run a little bit of captioning on it.

Q: Why WAI-illustrious-SDXL doesn't have Illustrious 2.0 or Illustrious 2.0 Stable?

A: It's because the creator doesn't want to put it into their model, they did explain this on the model page.

Each version of illustrious is trained on Danbooru up to certain dates. You'd have to look up those dates as I have forgotten what those dates are.

Section 2)

Section A) Basic Prompting

When prompting illustrious, you do not, run score tags like you do with pony as the model won't understand them
The more further a long a tag is in a prompt the more diluted it is, so you want your important stuff at the beginning of your prompt, as without layering Illustrious has, 248 tokens.
The bigger the prompt past the token capacity the more diluted the stuff at the end becomes diluted and forgotten. You'll want to rearrange your prompt in a way that has the positive quality tags at the end of your prompt, if you have a long prompt, and are prompting a OC or something that you don't want diluted.

For your positive quality tags in your prompt you can always start with:

masterpiece, best quality

Note: As we are running SDXL and Illustrious prompting, you won't need most things found on Pony.

For your negative quality tags in your prompt you can always start with:

worst quality, low quality,

Sometimes you don't need negative tags, sometimes you do and when you find your base negative prompt make sure to add on stuff you don't want in that image you are prompting. You may need to add to your negatives "loli, shota, child, aged_down" if you are prompting for mature content as most models as most illustrious models have a lot of that content due to Danbooru being littered with that type of content more so than the mature content, if you wonder why you've been getting younger looking images most of the time still, it is not full proof, putting only that in your negative so you may need to put "mature, mature <gender>" into your positive prompt, again it is not full-proofed. For Negative tags or Positive tags you want to add in check if they work on SDXL or go to danbooru to check if they exist, here is the link to the Danbooru Wiki Page: https://danbooru.donmai.us/wiki_pages/help:home

Section B) Basic Settings

Each Illustrious Checkpoint Models require different settings but I recommend to never go over 6 CFG as sometimes 6 CFG or above will overbake/overdo the image, more than what the model was trained on. If a model has no CFG for recommendations I suggest to start with using 3 CFG and work your way up to 6 to find what works for you, rather than just winging it and using settings you'd use on Pony or Regular SDXL which those two need higher CFG to function.

Say for one model the settings are:

Sampling Steps: 25-32
CFG Scale: 5.3
Sampler: Euler Normal

Another model might suggest:

Sampling Steps: 20-25
CFG Scale: 4
Sampler: Euler Normal

And another model might suggest something different and so on... just make sure to mess around with those settings to find what works for you. So it's best to find a few illustrious checkpoint models that are to your liking, or just stick to one checkpoint model. This goes for NoobAI, Pony and most other checkpoint models as well. Or...you can find settings that you prefer on most Illustrious models so you aren't stuck using 1 or 2 models.

For examples on how to prompt and arrange your prompt more organized and properly check out madbro666 or ecaj. They have some of the best organized prompts for Illustrious.

Section 3)

Section A) Basic Tag Info

Before we begin on Section 3A, most Illustrious models use Booru-Styled Prompting.

Some tags will not work due to them only being on NoobAI or Pony.
Some tags may not work because they are too small in number say an artist tag has 20 Images on Danbooru, you won't get that artists style and you'll have to resort to a artist style LoRA.
The link to figure out artist tags: https://danbooru.donmai.us/artists
Same thing goes for concepts, if a concept is too low in the amount of images on Danbooru it will not work. Unless the concept is also known by SDXL, as again Illustrious is a SDXL Finetune Model.
The link to figure out every other tag you can use: https://danbooru.donmai.us/tags

The best place to consult for tags when you are having trouble is Danbooru as it has most tags you can use.

Section B) How to use Tags

As for character tags you must use the order of the name that is shown on Danbooru.
Example: For Ash Ketchum you'd use Ash Ketchum or Ash_Ketchum, NOT Ketchum Ash due to it being a Japanese made anime.
We run tags how they are shown on Danbooru, not what we assume they are, for the most tag strength and effectiveness for the tag.
For color tags you will usually be able run <color> <tag> sometimes you have to use <color>_<tag>. Also for color tags, some colors might not be stated it knows but the model will know it, like Dark Blue. For Example if you prompt Dark Blue Sweater you will get a Dark Blue Sweater, unlike if you just prompt Blue Sweater it'll do whatever Blue color Bias the model has. Meaning all colors you can use in SDXL can be used in Illustrious, so there's a lot of colors you can use and can't use.
If a tag is like vex_(lol) since most sites and even most local places use ( ) as weights you will have to do vex_/(lol/) or vex /(lol/). So you're not weighting lol. The / in this case for most places, definitely not all, makes it so the ( ) is just the parenthesis in the tag and not being weighted, so you will 100% get Vex from league of legends.

Note: If you can't find the tag still, you can then ask someone else or as as tip, you can run autotagger like wd-eva02-large-tagger-v3 or the pixai v0.9 autotagger on the image with the tag you are looking for as it might autotag the tag you are looking for (I use it sometimes for that, you don't have to use the same autotagger) or you can run what everyone else runs to figure out what it is. Keep in mind everyone prefers different autotaggers, I just personally prefer wd-eva02-large-tagger-v3 and pixai v0.9 autotagger over the other autotaggers.

Remember to take this simple guide with a grain of salt and if I forgot anything or said some things wrong let me know what they are and how to fix them as I'd like to improve this guide. Before we begin the clip of a checkpoint model is the main knowledge it knows. This guide was made due to some things in Wolf999's guide being somewhat wrong for prompting, and most of the things being right or somewhat right. This guide is for Base Illustrious and may not be 100% accurate if you're running a merge model, finetune of illustrious, etc.

I'll also finish highlighting and coloring text, etc at a later date.

Arctenox's Simple Prompt Guide for Illustrious