Sign In

Chen Haoyu

190
1.3k
6
Verified:
SafeTensor
Type
LoRA
Stats
1,333
Reviews
Published
Jun 15, 2023
Base Model
SD 1.5
Training
Steps: 7,800
Hash
AutoV2
8795C36596
default creator card background decoration
losquit's Avatar
losquit
  1. So easy to train and publish LoRA and other kinds of models person these days. Yet the samples of many uploaders on Civitai are of obvious space to improve. 💤 They resemble not well enough. __*(  ̄皿 ̄)╯#

  2. What confuses me now is not only

    how to make the outputs resemble original person more, but also

    could the situation of so limited expressions, actions and scenes in Realistic base models (compared to Anime) be changed by some new checkpoint.

~~~~

Suggested weighted around 0.65. Notice "About this version" info.💁‍♀️

Lower your Denoising strength of Hires.fix especially on Chillout alike models.

Ture on extension "sd face editor" if necessary.

Refer to sample images and notice their unique negative prompts and sampling method of each checkpoint.

Suggested checkpoint KanPiro Mix, Chilled Remix v2, majic and so on.

(Welcome to feedback, review & give me more stars and red hearts!)

~~

Be kind to these lovely.

Legal & Risk

It is prohibited to use this model for commercial purposes and any scenarios of illegal acts and purposes.