PLEASE WRITE REVIEWS, IN THIS WAY I CAN IMPROVE MYSELF AND THE MODEL.
This is my personal model based on SD 1.5.
I have personally tested it creating more than 10000 unique images and it has always met my expectations. Positive and negative prompts affect the image in exactly the same way, so you'll have to play around with it!
Info:
-Close-up portraits, half-length photos and full-body photos give the best results but you can generate everything.
-The faces and in particular the eyes, if described properly, are surreal in how beautiful they are as this model generates beautiful faces by default, therefore positive prompts like "symmetrical face, perfect face, symmetrical eyes etc..." are useless and sometimes compromise the result, if necessary use negative prompts like "deformed, disfigured etc..." and the face will come out perfect.
-This model has the ability to also create landscape images (fantasy or not) but you will have to waste more time on it, as it is focused on"subjects".
-Be creative, describe the image in great detail, this guarantees a better overall result.
-NSFW have wonderful results both in txt2image and in img2img, if you want to transform a non-NSFW image into ???? and img2img does not give good results, use Inpaint and you will succeed 100%.
-Euler a, DPM++ 2M a Karras and DDIM seem to give the best results
Special thanks to sovereignrk and logoth for helping me with some new prompts!
1.0: OUT NOW (currently version)
2.0: OUT SOON (added many new concepts, fixed deformed hands, fixed deformed feet, fixed unreal eyes, fixed deformed fingers, fixed issue of unable to use embeddings, fixed duplicate subjects and many other little things)
STAY TUNED.
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
