Improved NSFW
Improved person diversity
Improved general coherence of generated images.













I spent many hours working on this model and merging many different models together to come up with this Eldritch abomination of a model. It has good realism and coherence and is overall a solid model.
Prompting is simple, just a simple sentence describing what you want. I recommend not using "quality descriptors", like UHD, 4k, photorealistic, as these don't meaningfully change the output image.
Here is an example of a prompt
Prompt: Portrait photo of a woman in the forest
Negative Prompt: BadDream, UnrealisticDream, desaturated, incoherent
There are also some descriptors that are fun to mess around with, like LSD/psychedelics at 1.2 weight, and other fun things I'll let you find on your own.
I recommend the following settings when using my model.
Sampler: DPM++ 2M SDE Karras or DPM++ 2S a Karras
2M SDE Karras is faster than 2S a Karras, both produce quality results.
Steps: 20-30
VAE: vae-ft-mse-840000-ema-pruned (not that important)
Clip Skip: 1-2 (If you don't know what this means, don't worry about it)
CFG scale: 5.5-7
I also highly suggest that you use the BadDream and UnrealisticDream negative embeddings with this model.
Please leave feedback on this model as to help me improve it. <3
Improved NSFW
Improved person diversity
Improved general coherence of generated images.
1. 轉载模型僅供學習與交流分享,其版權及最終解释權归原作者。
2. 模型原作者如需認領模型,請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
