Sign In

The Dragonsage Wildcards - Revision 1.4

19
345
5
Updated: Dec 12, 2024
toolfoodsmilewildcardswildcardsfw
Type
Wildcards
Stats
345
0
Reviews
Published
Mar 29, 2024
Base Model
SDXL 1.0
Trigger Words
____adult____
____child____
____dinner____
____meat____
____garnish____
____staples____
____vegetables____
____ethnic____
____expression____
____individual____
+11 more
Hash
AutoV2
01CBF45F89

πŸ˜πŸ‘©πŸ°πŸƒπŸŒ„β™£ The Dragonsage Wildcards

I often use the wildcards from this collection in my day-to-day prompts. In particular, the ethnic wildcard gets frequent use because it is one way to compensate for the predictable Eurocentric nature of the Stable Diffusion environment. I believe that ethnic diversity is a beautiful human quality that deserves to be β€Žillustrated and celebrated.

The ethnic wildcard is based on the Fooocus supplied nationality wildcard and inspired by PsionicRaccoon's Random Ethnicities wildcard:
https://civitai.com/models/232027/easy-random-wildcards-ethnicities-skin-tone-female-clothing-locations-poses-actions-hairstyles-and-more

This is the wildcard into which I have put the most work, but the ethnic wildcard represents only a fraction of the hundreds of ethnicities on the planet. When I put this wildcard in the prompt, I admit that I always anticipate the results with some excitement and it gives me great delight when I see the generated images.

Revision 1.4

- ethnic.txt - made one minor revision

NEW - melanin_flux.txt - 10 Afro skin colours, specifically designed for Flux models

Test images for each line in the melanin_flux wildcard are shown in the Gallery at the bottom of this page, but note that the order from lightest to darkest as well as the overall darkness of the set varies according to which UI and model you use. For example, in RuinedFooocus the whole set is darker and the order is changed compared to the images in the Gallery which were made in SimpleSDXL2.


Revision 1.3

- changes and additions noted in upper case

adult.txt - 2 entries: man, woman

child.txt - 2 entries: little boy, little girl

dinner.txt - a one line entry containing references to these 4 nested wildcards: meat.txt (60 entries), garnish.txt (30 entries), staples.txt (30 entries) and vegetables.txt (30 entries)

ethnic.txt - 366 entries - IMPROVED & EXPANDED

expression.txt - 124 facial expressions - IMPROVED

individual.txt - 10 entries: 5 life stages each for male and female persons - EXPANDED

melanin.txt - 14 Afro skin colours - IMPROVED

moves.txt - 24 ways for humans and animals to travel by foot - IMPROVED

nsfw.tst - 4 heavily weighted entries to control nudity in the negative prompt

person.txt - 4 entries: man, woman, little boy, little girl

place.txt - 64 entries: not only generic locations but seasons, times of day and weather

smile.txt - 124 ways to smile

Canadian Bonus Pack:

canadian.txt - 36 Canadian ethnicities - NEW

canadian_activity - 20 things that Canadians can do anywhere - NEW

canadian_city - 20 Canadian capitals and major cities - NEW

province_territory.txt - the 13 Canadian provinces and territories - NEW



Revision 1.1

- in ethnic.txt, corrected or improved some ethnic descriptors

- in ethnic.txt, added 30 ethnicities, mainly to improve coverage of ethnicities in China, India, Russia and the Pacific Islands

- by request, added expression.txt for facial expressions

- by request, added melanin.txt, to define Afro skin colour without specifying ethnicity. Because of the bias in the Stable Diffusion model, in effect this means that the ethnicity will default to Afro-American.

- revised person.txt to reduce the frequent tendency for Stable Diffusion to render a "girl" as a young woman - changed "girl" to "little girl". Also changed "boy" to "little boy" for consistency, although I have never seen a "boy" rendered as a young man.

- added dinner.txt and a set of associated nested wildcards: meat.txt, garnish.txt, staples.txt and vegetables.txt. Note that the categorizations of these foods are somewhat inconsistent: I was aiming for a workable display of a loaded dinner plate, rather than scientific accuracy.

- added an NSFW nudity control wildcard for the negative prompt, nsfw.txt. I have noticed that this wildcard will reduce the occurrence of revealing clothing, that technically is not actual nudity. When used, this one-liner wildcard should always be placed at the beginning of the negative prompt for maximum effectiveness, and it must be lower case.

Example Prompts

Prompt for the General Examples:
"full body long shot: a ethnic person moves through the place with a smile, 35mm lens, natural lighting"

Prompt for the Food & Expression Examples:
"a ethnic adult carries dinner on a platter with a expression, sharp background, clearly defined facial features, perfect hands"

Prompt for the Melanin Examples:
"a melanin woman walks through the city with a gentle smile, sharp background, smooth skin" (only 3 of 14 shown)

NOTE: The bold text represents a wildcard. Place two underscores together - i.e. a double-underscore - immediately before and immediately after each wildcard word

#fooocus #PureFooocusAI #stablediffusion #DavidSageAI