Sign In

Artist: gomzi

268
1.2k
10
Updated: Aug 30, 2023
styleanimegirlsstyle artstyle lora
03674-2206397291-1girl, bangs, blue_eyes, blurry, blurry_background, blurry_foreground, bokeh, bow, breasts, brown_hair, closed_mouth, coffee, co.png
03684-1712677350-1girl, long_hair, solo, christmas, blurry, snowing, snowflakes, blurry_background, looking_at_viewer, bangs, christmas_tree, dep.png
Images hidden due to mature content settings
Download (36.19 MB)
Verified:
SafeTensor
Type
LoRA
Stats
1,152
Reviews
(268)
Published
May 11, 2023
Base Model
SD 1.5
Training
Steps: 1,240
Epochs: 1
Trigger Words
1girl
Hash
AutoV2
5F8067500C
default creator card background decoration
Chanter's Avatar
Chanter
03674-2206397291-1girl, bangs, blue_eyes, blurry, blurry_background, blurry_foreground, bokeh, bow, breasts, brown_hair, closed_mouth, coffee, co.png
03684-1712677350-1girl, long_hair, solo, christmas, blurry, snowing, snowflakes, blurry_background, looking_at_viewer, bangs, christmas_tree, dep.png
Images hidden due to mature content settings

Gomzi Style LoRA

中文 | English

仿Gomzi的作画风格的LoRA模型。

权重建议值：0.8 (or 0.6~1.2)

基于模型：Anything-5RE

训练信息：共计620张原画（含裁剪图），迭代1240步，多分辨率迭代训练。

其他的一些建议：当权重为1时，头发颜色好像有点过拟合了，必要时可以着重强调头发颜色。

LoRA model with a painting style imitating Gomzi.

Recommended weight value: 0.8 (or ranging from 0.6 to 1.2).

Based on the model: Anything-5RE.

Training information: A total of 620 original paintings (including cropped images) were used for training, with 1,240 iterations using multi-resolution iteration training.

Other suggestions: When the weight value is set to 1, the hair color seems to be overfitted. When necessary, it is recommended to emphasize the hair color.