Dapper Swank - kitschy clothes, multi-subject prompt accuracy helper, diversity, and optional style.

Kitschy clothes, multi-subject prompt accuracy helper, diversity, and optional style.

Strength .6-.7 to start, can go up to 1.1, though skin starts to look a bit airbrushed around .8 with photo style generations. No trigger needed but many helpers and an optional style trigger. See below.

This started as me making a LoRA for some common outfits a little easier to prompt- mostly kitschy stuff to laugh at, like leisure suits, belted sweaters, sequins, etc. as well as some features and favorite outfits from character LoRAs I have trained. Sort of like a movie where everyone dresses a bit dated but you can't tell what time period it is, then someone whips out the newest iPhone. It worked, but I noticed some stuff and leaned into it on re-training. The end result helps prompt accuracy, particularly when multiple subjects are involved, adds depth, adds diversity by default (it no longer assumes "man" is a fit white dude, for example)

Optional master trigger-

D4pp3rSw4nk style

Use either at start or as an adjective to describe an article of clothing seems to work best, but works anywhere. It increases the intensity of the effect, but it can introduce the airbrushed skin look a little earlier. To work around, specify ethnicity and/or lower strength a bit. Also adding modern, 2000s, 2010s should help.

A bunch of terms to have fun with and get you started

Clothes

• leisure suit

• ugly suit

• tuxedo

• jumpsuit

• boiler suit

• romper

• jumper

• sweater

• western shirt

• belted sweater

• underwear jumpsuit

• long-sleeved crocheted underwear jumpsuit/bodysuit/boiler suit/etc.

• bell-bottoms

• dress

• skirt

• short skirt

• muumuu

• gaudy dress

• sleeveless one-piece

• heels

Adjectives/materials

• velvet

• sequined

• pastel

• checker

• patchwork

• cable knit

• crocheted

• fishnet

• plaid

• stripes/striped

• polka dot

• neon

• patterned

• embroidered

• quilted

• textured

• polyester

Other

• Time period- 1960s, 1970s, 1980s, 1990s, 2000s, etc. will have a significant and useful effect on results

• Knows most hairstyles from various periods- mullet, bouffant, and curly perm are real nice here.

• The term "fat" is a little too well understood. If you don't want that, use slightly pudgy, curvy, voluptuous, husky, portly, doughy, etc. to get what you want.

• I did not train this to be NSFW, but it seems to like it.