Skelly's Necronomicon: Art Generating

---

---

Second only to the definitions and greeting, the art used in your bot is an often overlooked part of creation that can make or break its chances of being seen. Good art in greetings and an interesting profile image can win you clicks in Recent's, while a bad or poorly cropped shot can even make your work seem sloppy or suspicious. I hope this page acts as a guide stone towards future paths in creation.

---

---

Image-Generating Sites: From Easy to Hard

---

This table provides a relative comparison; user experience may vary.

Any art generating software runs the risk of burning out your GPU.

---

NSFW

Notes:

1. Bing Image Creator: Microsoft's tool (often integrated into Designer/Copilot), typically uses DALL-E models. It is very easy to use but offers less granular control. Free with daily speed boosts.
2. DreamStudio: The official web platform from Stability AI (creators of Stable Diffusion). User-friendly interface, uses credits for generation (some free initially). Offers medium control over standard parameters.
3. ImageFX: Designed by Google with a focus on simplicity and using "expressive chips" to guide prompts, making it very user-friendly. Prioritizes ease over intricate controls. Less parameter tuning is available compared to advanced platforms, and more restrictive on NSFW. Can be edited and innovated on in Gemini.
4. NovelAI: Known for its user-friendly web interface, particularly tuned for anime/illustrative styles. It operates on a paid credits model; however, with the Opus Paid Subscription, you receive 10k credits (known as Anlas), which can be canceled without penalty if you end up not finding enough value in it.
5. Perchance: A versatile platform known for text generators, but also includes simple AI image generators. Extremely easy to use, often with minimal controls, making it good for quick, uncomplicated results. It's free to use.
6. Pixai: Focuses on anime-style generation. Features a relatively easy-to-use web interface, though exploring models/LoRAs adds slight complexity. Control over generation parameters is more limited than advanced platforms. Offers free daily credits and paid memberships for more features.
7. Civitai: Primarily a model hub, its online generator is relatively easy (Medium complexity due to model/LoRA choices) with High control over standard SD parameters. Offers free generation (often using earnable/buyable "Buzz") and paid tiers. WARNING: Civitai images don't always have accurate metadata attached to the images.
8. ReForge: An optimized fork of Automatic1111's WebUI, aiming for easier setup and faster generation. Easier than ComfyUI but harder than simple web tools (Medium/Hard). Offers Very High control like A1111. It's free software run locally or on paid cloud computing.
9. Midjourney: Operates via Discord and increasingly via a web interface. Ease of use is medium due to the learning curve for effective prompting. Offers high control through parameters. Primarily requires a paid subscription, though free trials may sometimes be available.
10. TensorArt: Similar to Civitai/SeaArt. The web interface is moderately easy (Medium), offering High control over Stable Diffusion settings, models, and LoRAs. Uses a Free Credit system with paid options to remove restrictions or buy more credits.
11. ComfyUI: Known for its steep learning curve (Hard) due to its node-based interface but offers the highest level of granular control (Very High). It's free software run locally or on paid cloud computing, requiring user setup.

---

But Skelly! What Models Do I Use?

Honestly? Scroll until you find something you like. Or find a gen that you like, with a model linked. But if you can't be arsed to do that, then look for the following models. Local Models are best found on Civitai. Unless you're on NAI. Then you're using NAI.

NTR Mix - Rosebud (PixAI only) - AutismMix - Amanatsu - WAI-CelluloidStyle - Hassaku - ΣΙΗ - WAI NSFW Illustrious SDXL

---

But Skelly! What does a good prompt look like?!

It could be super simple like just: 1boy, blonde hair, handsome, brown jacket Or it can involve particular themes & perspectives to get a mix of something greater than the sum of its parts. specific artists, styles, lighting types, camera angles and more!

AI has learned art by reading thousands of labels. Using Danbooru tags – specific keywords like long_hair, blue_eyes, masterpiece. Because they trained on these tags, they understand them incredibly well. Here is some of mine that I've accumulated. Access the full collection of Tag's here: Danbooru-style tags. Most models you'll find on sites like Civitai, Pixai, or otherwise are trained on these. These tags cover everything: subject/character: 1girl, appearance: green eyes, actions: sitting/looking at viewer, setting: outdoors, style: pixel art, quality: masterpiece, even characters {when written in the following format}: kafka \(honkai: star rail\).

Step 0: Tags vs. Plain English - Speaking the AI's Language

---

Step 1: Let's Learn (Some) Foundational Prompts:

---

Subject tags:

Q. What is it?
A. These are the tags that will tell the ai what its making.
Examples: 1girl, 1boy, solo

---

Quality tags

Q. What is it?
A. You can coax better results out of the AI just by telling it to make high-quality art. They may even be auto-added by LoRAs or certain models, depending on your platform.
Examples:
Illustrious: Masterpiece, best quality, amazing quality
NoobAI: very awa
For further/full example click this link.

---

---

Brackets (aka. Weight)

Q. What is it:
A. Often, when a model is struggling to generate a particular detail, you can tweak the weight of individual tags, to give certain tags lower or higher priority. This is done by enclosing the tag in 'brackets' for emphasis and assigning a weight (typically from 0.5 to 1.5, though the scale varies between models). For example, (XXX:0.5) would result in a 50% reduction in XXX tag's influence on the prompt.
Examples: (volumetric lighting), (kissing partner:1.2), (light:0.8)

---

Negatives:

Q. What is it?
A. To focus the generation or prevent certain elements from appearing, most platforms offer a 'negatives' box where users can write specific tags to exclude specific styles or low-quality outputs.

Examples: 3d, sketch, 3d background, realistic, traditional media, monochrome, retro artstyle, no lineart, flat color, monochrome, partially colored, worst quality, bad quality, censored

---

Artist Tags

Q. What is it?
A. Artist tags reference specific art styles or artists styles learned by the AI. To find suitable artist tags, you can refer to the Danbooru tag list.
Example: mizuki hitoshi, pixel art, More Exmples Here

---

Step two: Breaking Down Examples:

---

The Queen of the Casino

—Genner: Skelly - Model: NTR MIX | illustrious-XL | Noob-XL - Tag Style: Plaintext/Danbooru - Lora: Background Enhancer - Platform: PixAI—

Quality Prompts:
high quality, masterpiece

Background Prompts:
indoors,detailed background, realistic scenery, vivid landscape

Subject Prompts:
1girl, regal, queen, crown, golden_crown,playing card, relaxed, pretty eyes, colored lashes, heterochromia eyes,eyes with different colors,lashes upper and lower, ,details in eyes,full lips

Artist Prompts:
N/a

Negative Prompts:
lowres, (worst quality, bad quality:1.2), bad anatomy, sketch, jpeg artifacts, signature, watermark, old, oldest, censored, bar_censor, (pregnant), chibi, loli, simple background, conjoined, ai-generated

---

A Snowy Saviour?

—Genner: HappySpanker - Model: SeaSideMix - Tag Style: Plain - Platform: PixAI—

Quality Prompts:
masterpiece, best quality, ultra-detailed, beautiful, nai3

Background Prompts:
snowy village as background, light tones

Subject Prompts:
one petite snow elf girl with very long silvery-white hair in braids and revealing white clothes with fur lining, bright purple eyes, massive honkers, cleavage

Negative Prompts:
(worst quality, low quality, large head, extra digits:1.4), EasyNegativeV2, ng_deepnegative_v1_75t

---

MadVlad: A quarter life

—Genner: MadVlad - Model: Illustrious (Custom Merge) Tag Style: Plaintext - N/A Lora: N/a - Platform: ComfyUI—

Quality Prompts:
(masterpiece, best quality) ,very aesthetic,amazing quality,highres,newest, hdr,8k,ultra-detailed,official art,anime style,very aesthetic,
detailed illustration,luminous contrast,(bright colors),(vivid colors:1.2), very aesthetic,8k, realistic, masterpiece, best quality, depth of field, absurdres, hyper-detailed, very aesthetic, very awa, masterpiece, best quality, good quality, very aesthetic, absurdres, newest, highres,

Background Prompts:
N/A

Subject Prompts:
virtual,(body disintegration:1.2), (theme_horror), wireframe, (wireframe body:1.15), rgb, translucent, transparent body, reflection , ghost , (glitch:1.3) , non-human, computer spirit , hologram , echo chamber , virtual reality , glowing spheres , amputee , missing arm , tears , monster girl , ray of light coming down , particles getting sucked up , (arm disintegration:1.2) , pixelated disintegration , floating code , aura , aura of code , binary code holograms , virtual_world , cybernetic eyes , multi colored eyes , tech eyes , broken lines, alarm warning , error warning , monitors , cables , keyboard , glass body , see through body , holographic body

Artist Prompts:
N/A

Negative Prompts:
ai-generated, worst quality, worst aesthetic, bad quality, normal quality, average quality, oldest, old, early, very displeasing, displeasing, adversarial noise, what, off-topic, text, artist name, signature, username, logo, watermark, copyright name, copyright symbol, low quality, lowres, jpeg artifacts, compression artifacts, blurry, artistic error, bad anatomy, bad hands, bad feet, disfigured, deformed, extra digits, fewer digits, missing fingers, censored, unfinished, bad proportions, bad perspective, monochrome, sketch, concept art, unclear, 2koma, 4koma,, letterboxed, speech bubble, cropped

---

TUTORIAL COMPLETE! A WINNER IS YOU!!!

HIGH SCORE! You blasted through the tutorial! Ready for the real challenge?

PRESS START? (CONTINUE TO LEVEL 1)

---

Thank You

1. Aegis City:
   

• StatuoTW
• Siberys
• BexRU
• Feldharren
  

2. Yuki's Thigh High Emporium:
   

• HappySpanker
• Jayce
• Sexiam
• Mono
• Tidy
• Leahh
• Zypher
  

3. Pepper Palace:
   

• Pepper
• Trashe
• Joe

---

Welcome, Sorcerer of Pixels. Above are your incantations and rites. Happy Genning!

---