Doompypomp Style | Illustrious

Trained with 88 images

The style has no main trigger word so you don't have to worry about that BUT I did train in some characters to the style.

Android 18 is the most consistent character that comes out without a problem when it comes to her facial features but you'll have to use another LoRA if you want her clothes. Raven and Starfire work fine enough but I would still suggest using other characer or clothing LoRAs for them as their facial features and clothes didn't get trained in too well since there was little references of the clothes and not enough angles I used for their faces. Chi-Chi only got 4 images trained on her so only use with other character LoRAs based on her. A good way to get best of both is mix the character tags below for the style with actual character LoRAs based on them but slightly turn down the character LoRAs strength:

Chi-Chi from Dragon Ball:

chichiDP, 1girl, mature female, black eyes, black hair, hair bun, single hair bun, jewelry, earrings

Android 18 from Dragon Ball:

android18DP, 1girl, mature female, blue eyes, blonde hair, short hair, earrings, jewelry

Raven from Teen Titans:

ravenDP, 1girl, short hair, forehead jewel, purple hair, purple eyes, grey skin

Starfire from Teen Titans:

starfireDP, 1girl, red hair, long hair, green eyes, green sclera, orange skin, orange skinned female, silver bracers, silver armband

Also if you want the characters to be wet and sweaty just put in the tags "wet" and/or "sweat" without the quotation marks of course.

OTHER NOTES ABOUT GENERATING:

If you're generating with the Civitai Generator then try out Text to Image Hi-Res Fix option for good results.

Alright, now I'm getting the hang of how style LoRAs and training them work with this model. This one came out way better than the MadMark I trained a while back. If you're generating with this model you shouldn't have too many issues I think such as missing details since there's a bigger dataset compared to how I trained that one.

Even if you use Illustrious XL's base model you should be able to get a somewhat decent to good result depending how you use it and with what other LoRA models you combine with.

I will strongly recommend using Gehenna, Hassaku, WAI-NSFW or in general though and only play around with the base model Illustrious XL checkpoint if you're intrigued. I also have the Illustrious XL Personal Merge posted by dims2 as a suggestion but it can be hit or miss depending the type of character you're going for (I'd say if you're going for more anime looking characters or characters that have more details then it should do fine).

YOU DO NOT NEED TO PUT WATERMARK/SIGNATURE IN NEGATIVE TAGS SINCE I ONLY USED IMAGES THAT DIDN'T HAVE WATERMARKS/SIGNATURES IN THE FIRST PLACE FOR TRAINING!!

Anyway, Have fun with the LoRA