Type | |
Stats | 2,157 |
Reviews | (282) |
Published | Jun 4, 2024 |
Base Model | |
Hash | AutoV2 2C29B7256C |
WARNING! THIS MODEL IS HORNY! :3🦄
And you will like it! 🦄
V1.0 is more 3d with a bias towards realism, it can create anime style as well as 2.5d and 3d.
V2.0 is more horny version, it has more hentai, it makes hentai out of everything xD
VAE baked!
This model is the result of much experimentation.
I wanted to create something with which I could be creative and experiment further.
This model can create both flat anime style and 3d style with a bit of realism. It all depends on your positive and negative prompt.
I suggest using different Loras to get the result you want. The model is flexible and loves different Loras!
To give you an example, here are my favorite Loras that the model works best with:
Midjourney mimic. Art Style, super creative images, art, lots of detail.
Styles for Pony Diffusion Lots of different, cool styles that will enhance your image.
ExpressiveH (Hentai LoRa Style) My favorite Lora for expressive emotions ;3
And many, many more Loras!
Recomendations for use:
Use Hires Fix (1.5x-2x) that will greatly improve the final image!
Best for me - 4x_NMKD-Siax_200k
Cpil skip: 2
VAE: baked
CFG Scale: 5-7
Sampling method: I prefer DPM++ 2M Karras with 25-35 steps or Euler a with 40-50 steps.
DPM++ 2M - sharper and more detailed image
Euler a - softer and more gentle image
Resolutions: Any SDXL resolution, but i prefer 768x1280 and 832x1216
Use quality tags for nice anime result: score_9, score_8_up, score_7_up, score_6_up, score_5_up, source_anime, ... etc.
For more realistic style: masterpiece, absurdres, realism, realistic, 3d, artistic, cinematic lighting, ... etc.