LigmaSugmadi
patreon
Experiment: Kazusa images
🕑 Added 2024-10-12 22:41:44 +0000 UTCA little experiment I tried with local Loras to get some specific poses and then img2img with NAI to keep the artstyle to a noticable degree, added with Vibe Transfer from NAI. If you don't know what I'm talking about, what I'm trying to do is creating more poses at the cost of artstyle consistency. To make it looks this good, each image took about 20min to make. I'm still new at this so of course it took that long. In the future I plan to add something like this at the end of the set as a bonus. Trying to make many variations for these would take up so much time.
Comments
LigmaSugmadi
我有几种风格可以使用，它们已经在我的帖子中使用过了。人们似乎没有注意到这些变化，因为我确保风格不会偏离我通常发布的内容太远，所以最终可以称之为我的风格。关于人工智能的进步，我希望它能够得到改进，更容易使用，而不是更难。在目前的人工智能中，我或其他人仅凭提示就能想象出我想要的东西的程度是有限的。如果它进步了，用提示创建更多场景应该会更容易
朱里
你迄今为止的所有插图都非常精彩! 我想分享一下我的想法,我觉得这些改变可以改善你的作品。虽然最近你发布的作品没有发生任何重大转变，但由于人工智能技术的快速发展，我认为坚持一种风格是有风险的。放心大胆的尝试，我会一直在这里支持你。
Nautilus
Thank you for all your trial and error. I am supporting you in a small way.
mh03
Just wanted to let you know this is appreciated. A lot of creators in this space just post whatever uncohesive slop they can get away with, so it's nice to see some thought and experimentation going into sets.
