Only official art and anime frames used.








(So I'm going to start re-uploading my stuff.)
This v3 was trained only with official art and hundreds of frames from the trailer... which is basically the only animated source available. Couldn't find a version not compressed to oblivion, so this is about the best it gets for now.
Trigger: 'Haruka_ORAS'
Weight: [0.6 - 0.9]. Go high for art style fidelity (with less flexibility), and low for creative uses/poses. The model should remain consistent until around 0.65, but you may have to add the Basic Clothing Tags if elements start disappearing.
Basic Clothing Tags: 'red hairband', 'sleeveless red shirt', 'bike shorts under shorts', 'yellow fanny bag'.
Oh, and if you want to get rid of them, just add them into the negative prompt with a bunch of ((()))
Checkpoints used in examples: 'Anything V5' and 'UHD-23'
______
Example prompt:
1girl, haruka_oras, red hairband, solo, sun rays, (absurdres), perfect quality, (((seaside town, very windy, posing for viewer, modeling for viewer))) + [LoRA set to 0.75]
Only official art and anime frames used.
1. 转载模型仅供学习与交流分享,其版权及最终解释权归原作者。
2. 模型原作者如需认领模型,请通过官方渠道联系海艺AI工作人员进行认证。我们致力于保护每一位创作者的权益。 点击去认领
