T-ponynai3

这个版本是一个细节提升和语义理解提升的版本，在很多过往版本的prompt测试中都有了更好的效果，希望你们能喜欢这个新的细节微调后的版本，能在使用之前的prompt来生成时得到更好的效果，已经更贴近想要的效果的图像。This version is a version with improved details and semantic understanding, which has shown better results in many previous prompt tests. We hope you enjoy this new version with fine tuned details, which can achieve better results when using the previous prompt to generate images that are closer to the desired effect.

tPonynai3_v7.safetensors

v6.5

v6.1

v5.5

v5.1(weight optimized)

v4.1 (Optimized from v4)

v3.5

v3.0

v2.0

v1.0

<a rel="ugc noopener noreferrer nofollow" href="https://discord.gg/xnxRGpXJ" target="_blank">https://discord.gg/xnxRGpXJ</a> Here is the link to discord, there are two channels, Chinese and English451450868 小q群模型已经内置vae了，不需要额外添加vaeThe model already has included vae, there is no need to add additional vae最佳的出图策略是适中的分辨率开高清修复，而不是直接使用大分辨率直出The best generate strategy is to use high-fix at a moderate resolution, rather than directly using high-resolution direct outputpositive:(score_9,score_8_up,score_7_up,score_6_up,score_5_up,score_4_up)OR (score_9,score_8_up,score_7_up)负面可加：negative: (score_4,score_3,score_2,score_1),也可以加正常的nai系负面词，例如：negative: worst quality, bad hands, bad feet效果不佳时可以加anime，能调用更好的画风

T-ponynai3-v7-1.safetensors

Files with SHA256: a8f6faa7954285b36ccc339f38019d76089168465bbc6f56ef2c5b17a8e05d51