Sign In

【Portrait】 Vintage Gongbi Portrait Painting FLUX

26
129
30
10
Verified:
SafeTensor
Type
LoRA
Stats
129
30
Reviews
Published
Sep 17, 2024
Base Model
Flux.1 D
Trigger Words
gongbihua
Hash
AutoV2
9C0E5189A7
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

Hi~ I'm Juaner.

This is a model for generating beautiful gongbi-style portrait paintings of people. The color tone leans towards a vintage grayish-yellow style. The backgrounds are clean and simple. It excels at generating portraits of girls, but you can also try other people or animals.

Since these are portraits, the model may not perform as well for full-body images. It works better for generating half-body portraits.

In training, I found that if we don't have detailed description of appearance in caption, the lora of flux can easily generate average faces when training characters. it combines the average looks in the dataset to produce a fixed face. You can change hair color, but you can't change her appearance. This is the same even when training without captions.

So it looks like flux needs more detailed captions than SDXL. It is best not to use both data sets together

I don't know if there's any other model trainer who's had the same problem.

Recommended prompts:
Natural language + trigger words

For example: gongbihua,
A girl embodying the elegance of the Han Dynasty, with raven tresses meticulously coiled into an intricate updo, secured by exquisite jade hairpins, her almond-shaped eyes subtly lined with kohl, donning a crimson silk robe embroidered with majestic phoenix motifs, a delicate golden sash cinching her waist.

Recommended base model:
flux dev

Recommended weight: 0.8-1

Recommended size: 1024*1024

Please adjust settings based on the results.

Have fun, and I welcome everyone to share their generated images.

If you have any commercial licensing or custom model requirements, you can contact me at [email protected]

ღ( ´・ᴗ・` )

hi~我是卷儿。

这是一个用来生成好看人物肖像工笔画的模型,色调比较偏向于灰黄色复古风格,背景干净简单,比较擅长生成女孩,当然你也可以尝试其他人物或者动物。

因为是肖像画,所以它并不太擅长生成全身照,生成上半身图片效果会更好。

推荐prompt:

自然语言+触发词

比如:gongbihua,
A girl embodying the elegance of the Han Dynasty, with raven tresses meticulously coiled into an intricate updo, secured by exquisite jade hairpins, her almond-shaped eyes subtly lined with kohl, donning a crimson silk robe embroidered with majestic phoenix motifs, a delicate golden sash cinching her waist.

推荐底模:

flux dev

推荐权重:0.8-1

推荐尺寸:1024*1024

请根据结果灵活调整

have fun, 欢迎大家返图。

如果有商业授权或者模型定制需求可以联系: [email protected]

ღ( ´・ᴗ・` )比心