lbc_pony_simple

<p><span style="color: rgb(225, 60, 57); font-size: 22px;"><strong>Statement:</strong></span><span style="color: rgb(225, 60, 57); font-size: 22px;"> The model is based on the SDXL_pony_V6 model with two directional adjustments to obtain the base model!</p><p>Firstly, the art style is adjusted, the materials are sourced from my "lbc_Rebirth_Reborn_XL_v2" where I have output a large number of character sketches that are leaning more towards the 2.7D perspective, including "half-body" and "close-up" shots and a small portion of "full-body", with a proportion of 5:3:2. Hence, this model generally displays character half-body images by default unless your prompts include "full-body" or "close-up";</p><p>Secondly, I have added specific concepts for virtual character prompts to this model (mainly the virtual goddesses that I like, detailed character table will be provided below, and more will be continuously improved and added...) which means my model can directly output anime characters without the need for specific characters like Lora, this is my highlight! Of course, it's not limited to this, after adjustments, its generalization ability has also been greatly enhanced, so please cherish it!</p><p>Sidebar: My lbc_Rebirth_Reborn_XL_v2 actually cost me a lot of effort, but couldn't be promoted, I don't want to give up, so I still use it as always. With the blessing of the pony base model, the process of style adjustments went smoothly, over 5000 images, training over a month of intervals, this is how this model came to be!</p><p>Usage recommendations:</p><p>Pony uniformly uses aesthetic scores as quality words, so you should correctly fill in the corresponding aesthetic score words in the positive and negative prompt boxes, as follows:</p><table style="width: auto;"><tbody><tr><td colSpan="1" rowSpan="1" width="72">Positive Aesthetic Score:</td></tr><tr><td colSpan="1" rowSpan="1" width="72">score_9, score_8_up, score_7_up,</td></tr><tr><td colSpan="1" rowSpan="1" width="72">Negative Aesthetic Score:</td></tr><tr><td colSpan="1" rowSpan="1" width="72">score_6, score_5, score_4, source_pony,</td></tr></tbody></table><p>1. CLIP Termination Layer: 2</p><p>2. Sampler: DPM++ 2M, Euler a, DPM++ SDE</p><p>3. Recommended resolution: 2:3 or 9:16</p><table style="width: auto;"><tbody><tr><td colSpan="3" rowSpan="1" width="216">v1.0</td></tr><tr><td colSpan="1" rowSpan="1" width="72">Source</td><td colSpan="1" rowSpan="1" width="72">Character</td><td colSpan="1" rowSpan="1" width="72">Trigger Word</td></tr>

Files with SHA256: 42acd5c48aede25cd2392b1ee2fc8e289f0cd48a31053e391ed5a3eab42575a7