run Hyukev PonyXL for free on tensor.art
V2 Update
Better colors
Improved background
More versatile content
If you notice any issues, please contact me.
Recommended Settings
Sampler: Euler A / DPM++ 2M
Steps: 25 ~ 40
CFG scale: 5 ~ 7
Hires Upscaler: 4x-Ultrasharp
Prompts
Positive
score_9, score_8_up, score_7_up, Negative
score_5, score_4, source_pony, source_furry, censor, 3dSupport me ☕
https://ko-fi.com/hyukev
Description
FAQ
Comments (4)
2.0 is crazy. Most clear drawings without ADetailer I've ever seen.
My card is AMD so I couldn't efficiently inpaint, but this model is stably generating almost perfect eyes without inpaint at all. Usually with other models, either pony or SDXL, the eyes goes messy, but not here. Same for the body and hair.
The backgrounds also stepped up, highly detailed without weird, random strokes. The too bright problem from 1.0 also seems to be fixed.
Also shows better performance. With Hyu 1.0 or other ponys, under same configuration, one image for me usually takes 30~40 seconds. Wtih Hyu 2.0 it's 25~30 seconds.
Clearer, sharper, more detailed. Great update.
I appreciate that, I did my best! ❤️
Лучшие модели, которые я видел, среди аниме стиля. Все же, для аниме рисовки, больше подходит версия 1.0 особенно если наложить lora с персонажем. выглядит более естественным, а на версии 2.0 более рисованным, как будто это нарисовал художник. В целом супер, жду ещё и другие версии.
❤️