Pony Fantasy Knights - By HailoKnight
If you like what I do feel free to support me on Ko-Fi☕
Hello friends!
This LoRA is made to make amazing fantasy knight images for Pony models!
Things To know:
CHECK EXAMPLE PROMPT FOR USAGE!!
You should use the trigger word hkstyle
Use klip skip 2
Usage Tips :
The LoRA weight I have been using is 0.75-1, but feel free to try and experiment!
I have used ComfyUI when generated pictures and the settings im using are:
Sampler: dpmpp_2m
Scheduler: Karras
Steps: 30For A1111 enjoyers, I would use:
Sampler: DPM++ 2M Karras
Steps: 36Example prompts used:
Prompt:"score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, (beautiful demon woman, horns, wearing armor, holding pink glowing sword:1.2), dungeon in background, concept art, digital art, realistic, hkstyle"
Negative prompt:"text, watermark, low quality, worst quality, bad compostion, poor, low effort, child, childish, young, (3d, photo, hyperrealistic, rough sketch:1.1), censored, (derpibooru_p_low), furry, source_furry, source_comic"First image is with my LoRA and second is without my LoRA, same prompt and model.
Every picture is generated in 1024*1024 with no upscale
Recomended vae that I use is the official sdxl_vae
Description
FAQ
Comments (7)
Noob question, what makes Pony so different from SDXL models?
Good question I dont actually know either haha! What I have noticed is that Pony is better with certain poses and feels more natural.
"It was trained with a very high learning rate and it was captioned in a very special way with about 1.5 million images, enough to make the base model forget about everything it knew"
Training methods of different checkpoint,Trainers might choose different tags to describe some of the same things.Like score_9 and best quality,the mean of them might same,but ai doesn't think so.It can only understand the tag chosen by the trainer.
You will find that in different checkpoint, if you need an artist's style, you need different tags to describe it too.
Use different methods to train, more you train, The more different its understanding of tags……
Pony chose the unique tags to training.
This is what I understand.
technically most of xl was scrubbed, its whole new model. As to why its so popular its simple, as long as you are doing character related stuff (especially corn) it blows out of water every other model with its understanding and abilities to create different poses.
Unlike most models it understand both normal speech and boru tags, and you can mix both styles of prompts and it will understand them.
It also understands out of the box well over 1k characters from games/anime/movies/mangas some very obscure that you will not even find lore for them.
As far as human interaction, character posing goes its by far best model and with style loras made for it can produce extremely high quality results.
awesome LORA thank you
I hope we can get your other weapons for pony
Next one will be Witches and Wizards!!
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.