找回密码
 立即注册
查看: 1699|回复: 0

[传统艺术] [SD 1.5][Checkpoint]LOFI LOFI V1 fp16

[复制链接]

3869

主题

36

回帖

4331

积分

管理员

积分
4331
发表于 2023-6-7 21:37:14 | 显示全部楼层 |阅读模式

基础数据

样图
基本模型:
SD 1.5
小模型:
Checkpoint 
下载链接:
您必须是网站正式用户方可进行下载。

L.O.F.I: Limitless Originality Free from Interference

  • ? No special face alignment

  • ? Improve line details

  • ? Improve prompt understanding

Prompt suggestions

  • Since the text-encoder used is enough trained version, do not use a very high attention to control weight, which will cause some wrong drawing, it is recommended that all attention weights be no higher than 1.2

  • If there is no special composition requirement, there is no need for a lot of negative prompts, such as "missing hands", which may affect human body drawing, and DeepNegative is enough for negative prompts

  • It is strongly recommended to use hires.fix to generate, Recommended parameters:

    • (final output 512*768)

      Steps: 20, Sampler: Euler a, CFG scale: 7, Size: 256x384, Denoising strength: 0.75, Hires upscale: 2, Hires steps: 40, Hires upscaler: Latent (bicubic antialiased)

Most of the sample images are generated with hires.fix

Note: that if you use hires.fix, you may not be able to reproduce the image with the same set of parameters in webui, because hires.fix introduces double randomness

Model list

(To be verified) The weights of any byte in LOFI does not come from dreamlike model.

Model merge strategy

tldr: unet merge by grapeHentai(text-encoder,1 head layers) + hassanblend(IN-MID layers) + photorealistic(merged, OUT layers)

There are also some complicated merge modes, so I won’t write them here, and probably no one will read them?

(Welcome to my discord server to discuss if you are interested in model merging)

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则

Archiver|手机版|小黑屋|金房子

GMT+8, 2024-12-23 20:32 , Processed in 0.041939 second(s), 23 queries .

© 2023 金房子|AI发烧友社区

快速回复 返回顶部 返回列表