Welcome, One & All:

Third time a charm!

First off, i'd like to start by thanking Crody, they gave me excellent pointers, i've done many merges over the last year, but this model reminded me i'm still on a learning process and still a student. As such, their advices they have provided were really helpful, this model wouldn't have reached its full potential without them! Thank You

While i'm at it i would like thank Chumple for the testing, suggestions and advices, it's always a pleasure having support while trying to come up with a new model.

Thank you to mommymia to let me use their model!

About the Model:

Introducing RETRORDINARY, born from an experimentation i was making on the side to completely break away from my two other models styles.

Capable of rendering classic nostalgic retro pictures without the need of any loras, or if you just want to produce regular anime, semi-real even light painterly images, this model has you covered!

Guide:

Here's some tags to summon some cool retro-ness:

1) 1990s_style,
2) 1980s_(style),
3) 1980s city pop style,
4) 1990s aesthetic,
5) 1970s (style)
6) (80s anime style vintage),
7) 1980's cartoon style,
8) sets in the 80's,

1) retro art,
2) retro artstyle,

g-taste style, anime_style, manga_style,

Some cool styles:

1: (watercolor), (pastelcolor),
2: 8k, beautiful masterpieces, , thickly painted,solid color background, minimal composition,
3: soft lo-fi style,1980s anime style, pastel colors, simple shading,

Some japanese artists tags fitting the theme:

Koto Tsumi \(style\), (houjou raita), Terasawa Takeichi Style, kadoi aya style, studio ghibli inspired, Kuroinu Juu Style

(you can swapout 80s with 90s and vice-versa)

Quick example:

1990s (style),(anime coloring, anime screencap,official art:1), power \(chainsaw_man\)

Works best with DPM++2M at minimum at 20 steps, the sweet spot is between 25-50.

Euler a works too, might also get a bit crazy depending on your settings/scenes. Your mileage may vary.

Scheduler: Karras

Resolutions: 1216x832-832x1216 or 1024x1024

Guidance CFG: 3-8

The VAE has been baked in!

Known issues:

• Img2Img upscaling can be finnicky at times, i would recommend lowering the denoiser see if improves the quality. If you're using Adetailer, i would recommend using segment models rather than square detection ones (the mask detection stands out on the picture, especially if it's a dark setting/scene).

• Some pictures tend to be overdramatic...nothing i can do about that 😛!

Appreciation:

As always, thanks everyone for your continuous support throughout this last year, it's really encouraging and warming, i hope you'll love this model!

Thanks for your kind words, subs, tips, bids, shoutouts, generations, downloads, reviews, comments etc.... i really do appreciate it!

Stay healthy y'all! Take care of yourselves and your love ones! Peace Out!