Type | |
Stats | 1,710 140,182 |
Reviews | (302) |
Published | May 27, 2024 |
Base Model | |
Training | Steps: 2,886 Epochs: 6 |
Usage Tips | Clip Skip: 2 Strength: 1 |
Hash | AutoV2 64BBDA1D1E |
The new Illustrious model is fun but it's experimental and a little volatile. For more accurate '90s Disney-style gens, I still recommend the Pony version. (Read more below.)
***Notes on Pony vs. Illustrious version***
Whereas the Pony version can fairly accurately mimic the Disney Renaissance style, the Illustrious model brings more of a Disney-inspired flair to your animé gens, with flatter colors. The result is more Kimba the White Lion than The Lion King. The style is somewhat reminiscent of retro 1970/80s animé, like Lupin the Third. I'm excited to see what the community creates with it.
This LyCORIS (LoCon) aims to recreate the style of Disney's Renaissance period (1989-1999). It was trained on 480, high-resolution screencaps from nine films: The Little Mermaid (1989), Beauty and the Beast (1991), Aladdin (1992), The Lion King (1994), Pocahontas (1995), The Hunchback of Notre Dame (1996), Hercules (1997), Mulan (1998), Tarzan (1999).
Why This Model?
While many Disney models focus on the studio's modern output, this one taps into the nostalgic appeal of the '90s era, which holds a special place in my heart. Not only does this style resonate with me personally, but I also noticed a lack of such models for Pony. Shout out to thestanman for their SD1.5 Disney Renaissance style LoRA.
Guidelines for Use
Tags: No trigger word is required, but it benefits from the tag "retro artstyle." Additionally, "1990s (style)" or "1980s (style)" can be useful.
Example Images: Refer to the provided examples as a starting point for prompts and generation parameters.
Iconic Look: Use "blurry background" with "outdoors" to make the subject pop and achieve the iconic animation cel look of the period.
Furry Images: The model was trained on a lot of animals, particularly from The Lion King and Tarzan, adding a distinct Disney-fied flavor to furry/anthro images.
The model performs best when recreating subject matter similar to the source movies.