Please share you creations! I'm curious to see what you can achieve with my models!
Another model based on Pony XL. This model is intended to be similar to AutismMix, but with a slightly stronger anime feel and more "roundness", particularly on the flatter outlines and colors with a bit more contrast. It's made to be easy to use. In some cases, I find it's even more stable and predictable (consistent backgrounds, correct hands, etc).
It works particularly well on character designs and images with uniform outlines. Pure white / black background seems to be well supported. You can use it like AutismMix or Pony XL. If you are familiar with those models, the overall composition will mostly look the same.
This is a merge of AutismMix_DPO and LoRAs with a moderate weight.
Positive prompt example:
score_9, score_8_up, score_7_up, score_6_up, source anime, <whatever you want>Negative prompt example:
score_6, score_5, score_4, cartoon, ugly face, mutated hands, lowres, blurry face, bad ???????, bad proportions, deformed, deformed ???????, deformed fingers, censored, <whatever you want>Both versions produce similar results, but Lightning is much faster (only 8 steps). Classic can generate slightly more saturated images with more vivid colors, while Lightning is sometimes more accurate and works incredibly well with LoRAs (don't forget to check out my LoRAs that are 100% compatible with Smoothcuts).
Use this experimental version of Smoothcuts, a your own risk... or not: https://civitai.com/models/407624
Sampler: DPM++ SDE
Sampling steps: 8
CFG Scale: 1.5 - 2.0
Sampler: Euler a
Sampling steps: 30 - 40
CFG Scale: 6.0 - 7.0
This model comes with no warranty. Do not use this model for inappropriate purposes.
Tips:
Most potential small issues, e.g. with eyes can be very easily solved using inpainting/hires fix. There is no inpainting and hires fix used in showcase images.
Here's the lightning version of Smoothcuts v1.0 to generate your anime images at blazing speed!
You can normally expect the same level of quality from this version as from the original, but at a considerably reduced generation speed. Sometimes, the lightning version is even better in terms of prompt adhesion.
Recommended settings are:
Sampler: DPM++ SDE
Sampling steps: 8
CFG Scale: 1.5 - 2.0
1. Quyền đối với các mô hình được đăng lại thuộc về người sáng tạo ban đầu.
2. Người sáng tạo gốc muốn xác nhận mô hình vui lòng liên hệ nhân viên SeaArt AI qua kênh chính thức. Nhấp để xác nhận
