亚洲美女写真，，，小马模型，你懂的。我的第一次试水，但是我现在还不会写XL的提示词，慢慢摸索下。
我常用的参数：768X1344，euler a 30步，cfg 3-5（不要超过5） 高清放大的话，R-ESRGAN_4X+、4X-Ultrasharp、SwinIR_4X几个都差不多，重绘步数20、重绘幅度0.26.
内置VAE是跟平常默认的不一样的，可以看下对比图，左边是我的模型内置VAE，右边是普通模型通用的。对比度和色彩会有轻微区别。
欢迎大家返图，方便我的学习总结，尽快迭代新的版本！
Baidu Translate：
Asian beauty photoshoots, pony models, you know. My first trying, but I still can't write the prompt words for XL, so I'm slowly exploring.
The parameters I commonly use are 768X1344, Euler A 30 steps, cfg 3-5 (not exceeding 5). For high-definition magnification, R-ESRGAN_4X+, 4X Ultrasharp, SwinIR_4X are all similar, with a redraw step count of 20 and a redraw amplitude of 0.26
The built-in VAE is different from the usual default. You can see the comparison chart. On the left is my model with built-in VAE, and on the right is the general model. There may be slight differences in contrast and color.
Welcome everyone back to the picture, it will facilitate my learning summary and iterate the new version as soon as possible!
Description
FAQ
Comments (8)
本模型本来想发布在liblib的，但图片审核太严啦，封面的第一张图片我在liblib就无法过审，就发在这里了，否则就失去了小马模型都真正意义了，，，
封面图是由工作流生成的，好像无法看到参数信息
大佬加油！示例图很棒！
@fengyun321 感谢支持！随便跑的几张，刚玩小马，现在暂时还不会写提示词
赛博活佛，感谢
由于本地显卡的性能限制，我的图片都是fp8,8步极速生成的，模型的实际效果会更好些。
Due to the performance limitations of my local graphics card, all of my images are generated using FP8 and 8-step high-speed processing. The actual effect of the model will be better
When I generate images, only colorful images are generated.Do you know what could be the cause?
I have been using Comfyui and have not tested it on sdwebui. Additionally, please use the built-in VAE
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.