NoobaHoshi

All images use Text Prompt only, no Loras, ADetailer or HiResFix

Share your generations or opinions, I want to see if it would be worth to try another Merge method or not.

ABOUT THIS MODEL

I'm new to Models merge, and my goal in the past month was to merge a model with the styles and characters of NoobAi as intact as possible with more recent characters and more creative posing. In general, I want a model in which I would avoid using or limited amount of Loras.

The following models were used in this merger:

1. NoobAI-XL (NAI-XL) Epsilon-pred 1.1-Version as a base

2. Nova Orange XL v3 for its composition/poses

3. Raehoshi illust XL v3 for the extra characters data

* Check these amazing models that allowed this Merger to happen

I wish I had the recipe, but this was my first time doing Mergers and I did many different methods (around 30 mergers and even using Weight Block) and I didn't document what I did. 😅

This model is still not the ideal result I wanted, but is the closets I could get, if I can learn how to add concepts and characters without affecting all the styles of NoobAi I will try again, but for now this is satisfactory.

PERMISSIONS AND RESTRICTIONS:

✔Use only on CivitAI

✔Mergers allowed

✔Share art outside CivitAI

✔Credit me or name the model used

❌Use this model or Merges of this model outside of CivitAI

❌Sell Art

❌Sell Model

❌Claim it's your Art

HOW TO USE IT:

From my experimentation, avoiding quality tags like Masterpiece or Best Quality, it allows the Artist Styles to pop more, also BREAK can limit creativity at times, so play around and figure out the best results. Lora's use is uncertain, so please try and let me know.

#gender, character, series, character description, artist tag or style, quality tags, camera tags, theme or background tags, actions, clothing

EXAMPLE:

1girl, character, series, BREAK
anime screenshot, anime coloring, official art, very aesthetic, highly detailed, ultra detailed, absurdres, perfect composition, intricate details, elaborated designs, ultra-detailed Face, detailed eyes, dynamic angle, medium full shot, restaurant, coffee, snow, elbow rest on table, BREAK
elegant, ornate clothing, dress, high heels,

Negative Prompts:

6 fingers, sketch, worst quality, bad quality, bad hands, missing eyes, bad eyes, unfinished, unclear fingertips, twist, squiggly, incomplete, imperfect fingers, very displeasing, mess, approximate, sloppiness, artist name, error, extra digit, fewer digits, jpeg artifacts, signature, watermark, username, reference, disorganized colors, cropped, mutated, split view, manga, comic, twisted, censor, simple background, (artist name:1.2), tattoos, copyright notice, 

Settings I used:

Sampling method: Euler a
Sampling steps: 30
CFG Scale: 6 - 10

REGARDING AI:
My stance is that AI doesn't produce art, but it allows to materialize concepts/ideas, but I don't consider it art. So, IF I include prompts in the examples, I will most likely not include artist tags in the prompts.

Licenses of models used in the merger apply to this model too