Disclaimer: Ever since the teaser release I've seen a few LoRA of her popping up, some of them quite good to be honest but some others not really so I decided to jump into the Ronova boat too. Dataset for this one was 56 pictures at 1024x1024 to ensure better stability with a good balance between pictures from oficial trailer and the most official appearance accurate human made fan art I could find. Sadly the lower side of her dress is still quite a little inconsistent due to most media depicting her only in cowboy shots and rarely showing her legs or full body in detail.
Still this is far from perfect, I consider this an early release and will improve her once more official media surfaces over time.
Lastly, if you want renders to resemble more the official trailer use "anime screencap" tag, otherwise you can still render fanart/booru style art.
(Hires Fix and Adetailer use recommended for better quality pictures).
Recommended Weight: 1
General Tags:
Trigger: r0n0v4
Appearance: white hair, very long hair, hair intakes, white veil, hair ornament, hairpin, red wings, yellow eyes, star-shaped pupils, large breasts
Outfit: black dress, bodystocking, cleavage, detached collar, jewelry, elbow gloves, black gloves, see-through clothes, claws, black pantyhose, black footwear, shoes
Optional: anime screencap,
Note: In the official trailer her hair had a blondeish hue which blended into the LoRA, if you want her hair to look more white, add "blonde hair" into negatives to negate this yellow hue, also for some reason sometimes a red halo pops up, add "halo" into negatives to prevent this.
Optional Negatives: blonde hair, halo
Description
FAQ
Comments (4)
I didn't try it. But I'll and are you sure this lora can design for her pupils too? Most of them can't.
Yeah they should if you use the listed tags "star-shaped pupils". Usually when I work with characters like this I make sure to add additional training data for eyes sometimes I make the images myself (being an artist), dataset had around 7 eye closeups, some other pictures also had good quality shots of her face.
In any case I recommend you using Adetailer with an eyes model because Hiresfix alone normally is not enough to properly render good results with complex detail characters, all the preview pictures of my LoRA use both Hires fix and Adetailer with a Face pass and Eyes pass.
Added 3 more preview pictures with a face close-up, I will eventually refine this LoRA more and update it like I stated in the description because source media still lacks more consistent shots of her legs, I will make sure to include an even larger amount of training data for her eyes then because she has a lot of room for improvement so thanks to bring this to my attention, I'll add this detail to the top of the improvements list.
I like all Four Shades, but Ronova is definitely my fav out of all four.