The prompt text must start with: @null ghost.

提示词必须以：@null ghost.为开头
V2:
Version 2 uses the LOKR architecture and responds to words related to quality and the completion degree of the picture.

Recommended negative words (they are basically necessary; without negative words, the picture quality might be poor):

worst quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia, low quality, worst quality, blurry, bad anatomy, extra limbs, deformed, watermark, text, signature, bareness, artifacts, hands, copyrights name, jpeg_artifacts, scan_artifacts, bad hands, missing fingers, extra digit, fewer digits, artistic error, ye-pop, deviantart, logo, patreon logo

Sometimes, by changing the initial noise of the sampler from 1.0 to 0.95-0.90 and generating again, rare results can be obtained.

V2 has improved the horizontal aspect of the picture, bringing the overall quality to the upper limit that I can achieve. A stronger fitting effect may take some time. Try more prompts and combinations, and more detailed prompts can help you achieve the picture you envision. If your workflow consistently doesn't yield the result you want and feels abnormal, you can load the workflow by downloading sample images.
V2版本使用LOKR架构，并且对质量词和画面完成度的相关词语有响应。
推荐负面词（基本是必要的，没有负面词可能导致糟糕的画面）：
worst quality, low quality, score_1, score_2, score_3, blurry, jpeg artifacts, sepia, low quality, worst quality, blurry, bad anatomy, extra limbs, deformed, watermark, text, signature, bareness, artifacts, hands, copyrights name, jpeg_artifacts, scan_artifacts, bad hands, missing fingers, extra digit, fewer digits, artistic error, ye-pop, deviantart, logo, patreon logo
有时，将采样器的初始噪声由1.0更改为0.95-0.90,进行生成，会得到稀有的结果。

V2改进了画面水平，使得整体水准达到了我所能发挥的上限。更强的拟合效果也许得过一段时间。多多尝试提示词和组合，更详细的提示词有助于达到你所设想的画面，如果你的工作流一直得不出你想要的结果并感觉异常，你可以通过下载示例图片来加载工作流。

V1:
The prompt text must start with: @null ghost. This version has achieved most of the expected capabilities. I will train it again in the future to reach a higher level.

Note that this model is trained with the 40-layer architecture ANIMA2.9B, not the original 28-layer ANIMA base model. Mixing them together is likely to have no effect at all.

You should include the completion degree and style of the picture in the prompt, such as (draft, line drawing, doodle), etc., to control the quality of the generation. In this version, I forgot some things, resulting in it not having a fixed prompt to fix the highest quality style. This will be fixed in the next version.

If you use this model to generate images that you are proud of, please check this model into the usage list when uploading. This feedback is crucial for the improvement of my skills and the training of my model. Thank you very much.

This model is not allowed to be used in any way for all generation tasks that violate local laws and policies. Additionally, it is prohibited to generate with the purpose of "wash-up", "copycat", "creative plagiarism", "plagiarism", and "commercial use". When you download this model, all the results generated are your responsibility. Thank you for your understanding.
提示词必须以：@null ghost.为开头，这个版本达到了大部分所预想的能力，将来我会再次训练它，以达到更好的水准。

注意本模型为40层架构ANIMA2.9B训练，而不是原本的28层ANIMA base模型，如果混用，很可能不会发生任何作用。

你应该在提示词中加入完成度和画面风格，如（草稿丶线稿丶涂鸦）等来控制生成的质量，在这个版本中，我忘记了一些事情，导致它没有一个固定的提示词来固定生成最高的质量风格，这会在下一个版本中修复。

如果你使用了本模型来生成你所引以为傲的图片，请在上传时勾选此模型到使用列表中，这种反馈对于我和我训练模型的技巧改进中至关重要，非常感谢。

此模型不允许以任何方式使用于违反当地法规和政策的所有生成任务中，额外禁止以 洗图 冒仿 创意剽窃 和 商用 为目的的生成式推理，当你下载本模型时，产生的一切结果由你负责，感谢理解。