LeLo - LEGO LoRA for XL & SD1.5
Model description
LeLo stands for LEGO LoRA. It is a LoRA trained with over 900 images from the LEGO MiniFigures, BrickHeadz, and Creator themes. It provides a simulation of the LEGO design style.
Update:
The V2.1_SDXL1.0 version expands the training image dataset to 893 high-resolution images, further enhancing the model's drawing quality. Starting from this version, I will no longer provide updates for the corresponding SD1.5 version.
This version is released to celebrate the Lunar Year of the Dragon. Wishing everyone a happy holiday and prosperous fortune!
Usage:
Trigger words:
LEGO MiniFig, {prompt}: MiniFigures theme, suitable for human figures and anthropomorphic animal images.
LEGO BrickHeadz, {prompt}: BrickHeadz theme, suitable for human figures and anthropomorphic animal images.
LEGO Creator, {prompt}: Creator theme, widely applicable to objects, animals, plants, buildings, etc.
LoRA Weight: Between 0.6-1.0, recommended to use 0.8.
Denoising: Recommended to use 0.3.
Base model: The V2.1_SDXL1.0 version should be paired with the SDXL 1.0 series base models. For the SD 1.5 versions, it is recommended to use it in conjunction with the Realistic Vision series base models, but you can also try pairing it with other base models to explore different styles.
If you are interested in my other works, please visit my personal website at https://ai.chao.cool. There, I have listed my main achievements in Stable Diffusion LoRA and GPTs of ChatGPT.
LeLo 为 LEGO LoRA 简称。此 LoRA 模型使用了 900 多张 LEGO MiniFigures 系列,BrickHeadz 系列,和 Creator 系列产品图片训练而成,提供对乐高设计风格的模拟。
更新说明:
V2.1_SDXL1.0 版扩充训练图像数据集至893张高分辨图片,进一步提升模型绘图质量。从这个版本开始,我将不再提供对应的 SD1.5 版本更新。
此版本为庆祝农历龙年新年而发布,祝大家节日快乐,恭喜发财!
使用方法:
触发词(Trigger words):
LEGO MiniFig, {prompt}:人仔风格,适用于人物形象和拟人的动物形象。
LEGO BrickHeadz, {prompt}:方头仔风格,适用于人物形象和拟人的动物形象。
LEGO Creator, {prompt}: 乐高创意系列风格,广泛适用于物品、动物、植物、建筑等。
LoRA 权重(weight):0.6-1.0 之间,推荐使用 0.8
放大重绘幅度(Denoising):推荐使用 0.3
基础模型:V2.1_SDXL1.0 版请配合 SDXL 1.0 系列基础模型。SD 1.5 版本推荐与 Realistic Vision 系列基础模型配合使用,也可尝试与其他基础模型配合使用以探索不同的风格。
如果你的我其他作品感兴趣,请移步我的个人网站(https://ai.chao.cool/zh-cn/)。在那里列出了我在 Stable Diffusion LoRA 以及 ChatGPT 定制化 GPTs 方面的主要工作成果。
Trigger words
You should use LEGO MiniFig
, LEGO Creator
, LEGO BrickHeadz
to trigger the image generation.
Download model
Weights for this model are available in Safetensors format.
Download them in the Files & versions tab.
Use it with the 🧨 diffusers library
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained('stabilityai/stable-diffusion-xl-base-1.0', torch_dtype=torch.float16).to('cuda')
pipeline.load_lora_weights('lordjia/lelo-lego-lora-for-xl-sd1-5', weight_name='Lego_XL_v2.1.safetensors')
image = pipeline(' LEGO BrickHeadz,
In a field of golden wheat, a large doll with black short hair wears a yellow short-sleeve shirt topped with an orange suspender dress. White knee-high socks and black shoes adorn its feet, under a bare tree.').images[0]
For more details, including weighting, merging and fusing LoRAs, check the documentation on loading LoRAs in diffusers
- Downloads last month
- 6
Model tree for lordjia/lelo-lego-lora-for-xl-sd1-5
Base model
stabilityai/stable-diffusion-xl-base-1.0