All new lora build from scratch
I trained this lora to recognize all aspects of her like clothing, hairstyle, and even that the images I trained it on are CGI Images so that SDXL knows when you remove that and put it in the negatives you want something else so let's say a realistic photograph. I will explain each aspect.
Use the following words in your prompt to generate the desired object
Subject:
Jill V (this is also the main trigger word)
Clothing:
wearing a blue tanktop with a white layer underneath
wearing ripped jeans or wearing black jeans
wearing black gloves or wearing black finger less gloves
wearing combat boots, wearing shoulder holster (this is only the shoulder straps for upperbody shots if you want the actual gun holster add gun holster doesn't always work though)
Wearing necklace
Holding a gun, aiming (This does sometimes create weird hands unfortunately)
Hairstyle:
Side parted bob hairstyle
dark brunette
Creating a specific style:
Game Jill CGI:
add CGI, unrealistic, CGI animation still to your prompt and use SDXL Base or a CGI based model
Realistic
Add CGI, unrealistic, CGI animation still to your negatives
Use SDXL Base
Extreme realistic
Add CGI, unrealistic, CGI animation still to your negatives
Use a realistic model like BATCH CO REALISM (XL) or any other realism model
Lower the lora weight
Trigger word: V1.0 JillVal2 V2.0 JillRE woman
Strength :1 works fine but if you want realistic you have to play a bit with the strength of the LORA and the trigger word. Seems to be reacting to different models differently but overal generates a pretty accurate Jill Valentine.