Images(half of it) have inside comfyui workflow even if civitai don't say it. Just open image in new tab->save->move it to comfyui by drag&drop.(use comfyui manager for dl missed nodes, and maybe it not dl this one: https://github.com/farizrifqi/ComfyUI-Image-Saver)

LoRA size is big because much characters, if i train it smaller or resize now it lost much details and consistency on BG like here.

"Train data" dl option is trigger words/wildcards.

if someone interesting in commision PonyXL LoRAs - price is 15$/character for multicharacter LoRAs(3+), and 20$/character for solo character(15$ too if you request more than 3 charas from one title in one order). Write PM here or in discord.

I recommend multicharacter, it works great on PonyXL and you get ability to made images with 2 or more characters by prompt without problems with regional prompter technics(and even with it better than use 2 different LoRAs).

this LoRA is example of quality you can test are you can control it and get good result.

Train LoRa is time and resource expensive, you can support my future works here:

❤️Boosty Donation link

old text:

Astarotte no omocha!

    <LINK>       <LINK>        <LINK>        <LINK>       <LINK>

    <LINK>       <LINK>        <LINK>        <LINK>      

    <LINK>       <LINK>        <LINK>        <LINK>       <LINK>

    <LINK>       <LINK>        <LINK>      

    <LINK>       <LINK>        <LINK>        <LINK>     

    <LINK>       <LINK>        <LINK>     

    <LINK>       <LINK>        <LINK>        <LINK>     

Resize, rename to usable format(<title>_<trigger word>), compile to one civit page, and made this page look fine.

It's same as old one, but much less size with little lost in quality(yes older is seems little better in details, but not so much).

all model better even in details, old ones unpublish for decrease count of bad working models on my account.