|
--- |
|
license: other |
|
license_name: bespoke-lora-trained-license |
|
license_link: https://multimodal.art/civitai-licenses?allowNoCredit=True&allowCommercialUse=Rent&allowDerivatives=True&allowDifferentLicense=True |
|
tags: |
|
- text-to-image |
|
- stable-diffusion |
|
- lora |
|
- diffusers |
|
- template:sd-lora |
|
- chinese |
|
- realism |
|
- asian |
|
- elderly |
|
- style |
|
- children |
|
- middle-age |
|
|
|
base_model: stabilityai/stable-diffusion-xl-base-1.0 |
|
instance_prompt: |
|
widget: |
|
- text: 'a middle-aged man in suit in front of a building, half body ' |
|
output: |
|
url: >- |
|
2497782.jpeg |
|
- text: 'photograph of a female high school student in class room, half body' |
|
output: |
|
url: >- |
|
2496560.jpeg |
|
- text: 'portrait photograph of an old man on street, half body' |
|
output: |
|
url: >- |
|
2496563.jpeg |
|
- text: 'a young man on street, half body' |
|
output: |
|
url: >- |
|
2496564.jpeg |
|
- text: 'a young woman on street, half body' |
|
output: |
|
url: >- |
|
2496562.jpeg |
|
|
|
--- |
|
|
|
# 当代华人 Contemporary Chinese for XL & SD1.5 |
|
|
|
<Gallery /> |
|
|
|
|
|
|
|
([CivitAI](https://civitai.com/models/81771)) |
|
|
|
## Model description |
|
|
|
<p>尝试优化基础模型中缺乏华人形象的问题，使用 500 张人物环境肖像照，典型职业肖像照和环境照训练，力图覆盖不同性别和年龄段。</p><p>使用时无需 Trigger Words，在 SD 1.5 下权重一般不超过 0.8，在 SDXL 1.0 下权重可提高到 1.0。</p><p><strong><span style="color:rgb(253, 126, 20)">注意：</span>easynegative</strong> 和 <strong>ng_deepnegative_v1_75t</strong> 两个 Embeddings 会影响此 LoRA 效果，请避免同时使用。</p><p></p><p><strong><span style="color:rgb(253, 126, 20)">V2.0 for SDXL 1.0 更新：</span></strong></p><p>SDXL 1.0 对亚洲人形象的支持，相对于 SD 1.5 已经有了很大的提升。 我仍然针对 SDXL 训练了一个版本，在基础模型表现不佳的情况下可以尝试使用此 LoRA。</p><p>以下的说明主要针对 SD 1.5 版本。</p><p><strong><span style="color:rgb(253, 126, 20)">V2.0 for SD 1.5 更新：</span></strong></p><ol><li><p>以 fp16 精度保存，以提供对早期显卡的支持。</p></li><li><p>基于 Realistic Vision v2.0 训练，大幅提高 epochs 到 80，大幅提高了人物形象和场景的真实性。</p></li><li><p>由于真实性的提高，可能在一些情况下，美学效果有所降低。如需要可继续使用 V1.2 版。</p><p></p></li></ol><p>以下是基于 Realistic Vision v2.0 进行的 4 组对比，展示基础模型原生亚洲人（Asian），华人（Chinese）与本 LoRA 在权重 0.8 下生成的华人形象的区别：</p><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/434bfa58-4be2-407c-a3a7-f17489f587b2/width=525/434bfa58-4be2-407c-a3a7-f17489f587b2.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/27b09070-d7a0-4161-9366-91128a775230/width=525/27b09070-d7a0-4161-9366-91128a775230.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/240c48fb-118b-4032-8734-8f159aea26d3/width=525/240c48fb-118b-4032-8734-8f159aea26d3.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/a49fe3ba-17c9-43a1-b004-ad7238c6ed50/width=525/a49fe3ba-17c9-43a1-b004-ad7238c6ed50.jpeg" /></p><p>此 LoRA 可与主要的基础模型配合使用。以下是基于 5 个常用基础模型（Realistic Vision v4.0，ReV Animated v1.2.2，Deliberate v2，ChilloutMix，Disney Pixar Cartoon type B v1.0）的应用效果对比：</p><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/789d0999-2453-4793-9940-cf112c5a2ea4/width=525/789d0999-2453-4793-9940-cf112c5a2ea4.jpeg" /></p><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/b347c15a-0bfc-497e-9d8e-be60a8b25934/width=525/b347c15a-0bfc-497e-9d8e-be60a8b25934.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/1307c24b-8ad2-4313-88b2-95bea6997095/width=525/1307c24b-8ad2-4313-88b2-95bea6997095.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/6dd2cffe-574a-48e9-969b-72f337cc5633/width=525/6dd2cffe-574a-48e9-969b-72f337cc5633.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/df70a76d-5b40-49d7-8006-cf9bb6b2a41f/width=525/df70a76d-5b40-49d7-8006-cf9bb6b2a41f.jpeg" /><br /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/7a52e344-ecfe-4c4a-871a-5cfb77999de8/width=525/7a52e344-ecfe-4c4a-871a-5cfb77999de8.jpeg" /><br />如果我的工作对你有帮助，欢迎留下你的评价和评论，也请试用我的其他 LoRA 作品。你的支持对我非常重要。</p><p></p><p>最后埋了一下快递小哥的彩蛋，请自行挖掘😀</p><p><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f11dd8cc-f1b1-4edb-a8e7-d1e89b3b992a/width=525/f11dd8cc-f1b1-4edb-a8e7-d1e89b3b992a.jpeg" /></p> |
|
|
|
|
|
|
|
## Download model |
|
|
|
Weights for this model are available in Safetensors format. |
|
|
|
[Download](/lordjia/dang-dai-hua-ren-contemporary-chinese-for-xl-sd1-5/tree/main) them in the Files & versions tab. |
|
|
|
## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers) |
|
|
|
```py |
|
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/dang-dai-hua-ren-contemporary-chinese-for-xl-sd1-5', weight_name='CC_v2.0_XL.safetensors') |
|
image = pipeline('a young woman on street, half body').images[0] |
|
``` |
|
|
|
For more details, including weighting, merging and fusing LoRAs, check the [documentation on loading LoRAs in diffusers](https://huggingface.co/docs/diffusers/main/en/using-diffusers/loading_adapters) |
|
|
|
|
