模型

模型有兩個(gè)，分別是 stable-diffusion-xl-base-1.0、stable-diffusion-xl-refiner-1.0。

base 模型是用來(lái)做文生圖，refiner 模型是用來(lái)做圖生圖的。

SDXL 模型之 base、refiner 和 VAE_云水木石的博客-CSDN博客

分辨率

默認(rèn)是1024*1024，或者像素?cái)?shù)總數(shù)不大于1024*1024，且接近這個(gè)數(shù)

# 768 x 1344: Vertical (9:16)
# 916 x 1144: Portrait (4:5)
# 1024 x 1024: square 1:1
# 1182 x 886: Photo (4:3)
# 1254 x 836: Landscape (3:2)
# 1366 x 768: Widescreen (16:9)
# 1564 x 670: Cinematic (21:9)

Samplers

DPM++ 2M Karras?or?DPM++ 2S a Karras

VAE

參考：SDXL 模型之 base、refiner 和 VAE_云水木石的博客-CSDN博客

PROMPT

常用

模板

PROMPT

breathtaking {prompt} . award-winning, professional, highly detailed

NEGATIVE PROMPT

anime, cartoon, graphic, text, painting, crayon, graphite, abstract glitch, blurry

?應(yīng)用

PROMPT

breathtaking selfie photograph of astronaut floating in space, earth in the background. award-winning, professional, highly detailed

NEGATIVE PROMPT

anime, cartoon, graphic, text, painting, crayon, graphite, abstract glitch, blurry

LORA

??

I Made Stable Diffusion XL Smarter by Finetuning it on Bad AI-Generated Images | Max Woolf's Blog

參考：

https://aituts.com/sdxl-prompts/

prompt

AI繪圖提示詞Stable Diffusion Prompt 筆記_linzhiji的博客-CSDN博客

https://github.com/Mr-jiangrj/Prompt?