https://civitai.com/models/438024?modelVersionId=487931
Lieri Bishop is on PonyXL now!
SD 1.5 animefull version and Naomi Evans will be soon.
PonyDiffusion v6 (XL) settings:
Style settings for PonyXL (very important):
score_9, score_8_up, score_7_up, score_6_up, score_5_up, uncensored, BREAK
Use Autismmix Pony instead of pure PonyDiffusion v6 XL because it much more stable in styles.
Personally I like lycoris girl next door ("gnd" in my prompts). It does picture much better even with 0.4 weights.
Also there are other good style LoRAs.
For checkpoint mixes like mfcg styles are not important.
Use 3D in positive or negative for 3D arts / anime arts. For 2.5D LoRA don't use 3D in positive. It is very strong tag.
Source_anime is very situative. It can do worse (cause oversaturation). I think it is better without LoRA.
Negative prompt for PonyXL (recommended):
censored, blurry, head out of frame, watermark, jpeg artifacts, muscular female
Also add in negative annoying objects.
Best tags order for PonyXL:
-
score
-
Sentence (for example, "A girl is standing with crossed arms and smiles")
-
tags for concepts, camera, face, body, clothes, background, etc. Order is the same as for 1.5.
Preview images can help.
PonyXL character tags:
Lieri Bishop, 1girl,
purple hair, long hair, green eyes, bangs
military uniform, white miniskirt, black thighhighs, lace trim, garter straps, garter belt, ankle boots,
optional: black bra, maid headwear, maid apron, micro bikini
Also you can try other tags (police girl, school uniform, etc). Preview can help.
Advices for upscaling / fixing:
use highres.fix (If you are sure that your prompt is easy for SD); I prefer x4_Fatal_anime_500000_G, foolhardy_remacri with 2x upscale. 0.35 denoising and 15-20 Hires fix is enough.
Do not use "latent" upscaler because it works worse with denoisning < 0.6 but you don't want so high denoising stength because it changes your image instead of adding details and causing mutations.
use extention after detailer for better faces or img2img inpaint (only mask) for faces or eyes with 0.25 denoising.
use SD Upscale / Ultimate SD Upscale script in img2img after highres fix if you want 4K or more resolution with adding details (don't use Extras for it).
Use Photoshop / Photopea extension for fixing hands, textures and other.
I will be glad to receive feedback from you and a demonstration of your art with my LoRA!
Don't forget to like LoRA model :3
https://tensor.art/u/604375505701151021
