### PlumpShapeXL

(This is my first LoRA, please be nice to me!)

*WARNING*

The following description was partially crafted by AI. The claims herein may be sensationalized but are essentially accurate. Proceed once you have acquired an appropriately sized grain of salt.

PlumpShapeXL captures a bold, clean, graphic art aesthetic with uniform ink lines and smooth digital gradients. Because of clean dataset curation, it acts as a triggerless LoRA. It applies its unique body proportions and line weight to any character or gender while respecting your custom text prompts.

🏷️ Trigger Word & Style Reinforcement

PlumpShapeXL is triggerless. You do not need to type a specific keyword to activate the LoRA. It will automatically apply its unique shapes and smooth line art to your generation just by being active.

However, if you want to push the aesthetic even further and strongly reinforce the look, you can manually add all or some of the core style tags used during dataset training to your text prompt as needed (or not):

• For Outline Depth: outline, white_outline, thick_lineart, black_outlines

• For Cartoon Rendering: cel_shading, flat_color, cartoon, 2d_art

### 📐 Body Type & Proportion Customization

PlumpShapeXL has a distinct architectural map hardcoded into its DNA. By default, it introduces an ultra-wide, voluminous hip-and-thigh structure and a voluptuous body. This is a bold graphic choice that weights the character's silhouette downward, replacing standard, slender anime frames with a thick, solid shape.

Because generic body-mass descriptors were completely purged during dataset curation, this structure is incredibly flexible and can be modified in two ways:

• 🧬 Universal Adaptation: It applies its signature curves to any** character, franchise, or gender—even non-human or anthropomorphic figures*—without causing anatomical glitches, gender distortion or compromising distinct character features. (*In limited testing) (**most?-ish?)

• 🥞 Proportion Stacking: The model acts as a stable foundation for custom prompting. If you want an even thicker, more exaggerated body type, you can manually stack size and weight tags in your text prompt. PlumpShapeXL will scale up naturally, expanding the proportions smoothly without sacrificing the clean, bold lineart or breaking the art style. (see BethXL in the samples)

It leans towards a cartoon style but performs fine on more realistic checkpoints with higher strength values, it is pretty flexible in either direction. If it is influencing the style too much I found that, in testing, a lower strength will start to drop the Toon style while mostly retaining the body proportions of the LoRA. (or you could crank it up, I'm not your dad.)

Most samples were generated with Prefect illustrious XL and all were generated with Hires Fix and ADetailer for faces. (the semi-real sample was made with Pearly Opal semi-real).

I have included some comparison images with the samples so you can see the difference the lora makes. The corresponding images were generated with the same prompt and settings with the LoRA being the only difference.

🔧 RECOMMENDED SETTINGS:

• Base Model: Illustrious-XL or compatible checkpoints.

• LoRA Weight: 0.8 - 1.0 for Anime/Cartoon.

• CFG Scale: 5.0

• LoRA Weight: 1.2 - 1.6 for Semi-real.

• CFG Scale: 5.0

You can, of course, raise or lower the strength or CFG at your discretion, I found it was pretty flexible in testing. These are just the settings I used to make the samples,

Like i said at the top I'm relatively new to this so I'd be happy to hear your feedback or answer any question that I can. I hope you find use for PlumpShapeXL.