-This is my Second XL Chara Lora (Slightly improved version)-
Base model : Animagine XL V3
Recommend a weight : 0.65 ~ 0.85
Recommend Models : Animagine XL V3
Recommend Prompts (Clothes) :cropped blue jacket,black pubic tattoo,thighhighs, gloves,boots,garter straps,thong,underwear,etc...
Recommend Prompts (Chara) :kcl,1girl,
※Since the tag "kcl" aggregates tag elements such as hair style, hair color, etc., you probably do not need to bother setting it.
These LoRa are not compatible with Animagine v3.1.
You should use this LoRa in conjunction with Animagine v3.0.
-Major Improved areas-
①About the tattoo area
②About the jacket part
③The part about tagging
-Review(for personal reflection)-
As for my personal level of completion, I feel that I was able to raise it to around 85% from the previous level of around 80%.
-Dear AnimagineXL model author-
I am very sympathetic to the sadness of many people who throw critical and negative words at us in today's world. I can understand it very much because I have experienced the same thing in the real world. But remember that there are more people who support, respect and admire you than those diminutive people!
Of course I am one of those who adore you.
I hope that whenever you feel better, you will resume your wonderful creative work!
Note : This is an experimental Lora to check its versatility, I may create a new version to increase its versatility.(Mainly about learning materials and tags)
I would like to see the work of everyone using this Lora, whether normal or NSFW, so please post a variety!
The electricity bill is so terrible that I will weep and rejoice if you could grace me with a cup of coffee. ⇒https://ko-fi.com/x_milaidou_x
Description
FAQ
Comments (3)
Great Lora, so impressed I was with the outputs I installed ComfyUI and SDXL models just for this ! It renders great but a bit hard to get some positions/poses right only from prompts, is there a way for my 6GB GPU to make more prompt accurate images? Thank you for the work on this and hope you will update in the future! If I can also test it in other ways you would like to see, let me know, I really like this !
Thank you for your comment.
For SDXL operation with 6GB, perhaps ComfyUI can be used with the least amount of VRAM used, but another idea would be to use "Forged UI"?
https://github.com/lllyasviel/stable-diffusion-webui-forge/releases
I think the "Pony" version allows for more expression than the "Animagine" version. The only drawback, however, is that the background is simplified.
The "Pony" and "Animagine" versions have their own advantages and disadvantages, but the "Refiner" function will allow you to express the best of both worlds to some extent.
Another alternative would be to use a free AI image generation site for output.
It might be a good idea to have the model data created sequentially uploaded to a well-known site.
(It is a pain in the ass to deal with the fact that some people are uploading my model data without my permission, impersonating me...)
I guess the only thing left to do is to use some quality-improving Lora for a little more expressiveness?
I have applied several quality improvement Lora to some of the images I have posted, so I hope this will help.
As for the ones I am getting from outside of Civitai, they are the following two
I am listing this information for my own recall purposes as well.
https://huggingface.co/2vXpSwA7/iroiro-lora/blob/main/sdxl/sdxl-flat.safetensors