Sign In

Holiday Extravaganza of the Stars 2024

11
41
8
2
Verified:
SafeTensor
Type
LoRA
Stats
41
8
31
Reviews
Published
Dec 21, 2024
Base Model
Flux.1 D
Training
Steps: 9,720
Epochs: 10
Usage Tips
Strength: 0.7
Hash
AutoV2
AC85C21FFD
The FLUX.1 [dev] Model is licensed by Black Forest Labs. Inc. under the FLUX.1 [dev] Non-Commercial License. Copyright Black Forest Labs. Inc.
IN NO EVENT SHALL BLACK FOREST LABS, INC. BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH USE OF THIS MODEL.

This is insane, and insanely fun. Many styles and concepts.

TL;DR simple use

strength around .7 to start, go higher for a more vintage style, lower for more realistic, talk about Christmas/Hanukkah/holidays and common related topics. Will make pictures that look like portraits from a holiday card, packaging, studio portraits, comedic insanity, etc. No trigger in particular needed (but a ton of fun optional ones- see below).

Dumb short prompts work best for comedic results, prompt as usual and a little lower strength for more serious/realistic.

The Real Fun

  • Photo, photograph, portrait, illustration etc. are good starting points

  • A holiday card, a card, a Christmas card, a Hanukkah card, A slapstick holiday card, etc. Tell it what color border or background if you don’t like what it is defaulting to with your prompt.

  • Sparkling

  • Metallic sheen

  • Vintage/retro

  • Slapstick

  • With sparkling blue-to-sepia ombre burst vignette effect (change the colors, the effect varies with the rest of the prompt and strength. You can swap out sparkling, ombre, vignette, reorder to change effect, etc.)

  • It loves grain

  • Collage

Characters, Traditions, and Concepts

I have corrected holiday lore and made it Biblically accurate per both Testaments and the Torah, as well as the various books and chapters that have been revised, scrubbed, or omitted over the centuries. Some are more consistent than others, and all are consistently insane.

The characters are a type of subject/outfit rather than an exact character. Names used of real people are placeholders and do not give any likeness.

I will share some basics (Some do exactly what you think they will, some do the opposite, some are just ridiculous and a little trolly at times). See examples for proper usage. This is barely the surface :)

  • Santa

  • Krampus- there are a few variations from a guy with horns to a wolf horn beast thing.

  • Krampusette

  • Reindeer

  • The lone wiseman bard

  • Blueberry walrus

  • Rudolph the red nosed reindeer

  • Angel Gabriel

  • Elf/elves and some modifiers (inconsistent in a good way- Western elves, Elves from Space Force, Old elves, Crust elves, etc. If it tries to make an elf a kid, call it old or ancient or something they has a very strong “not a kid” connotation)

  • Eastern European relatives

  • Batman-Christ

  • Tinsel slug

  • Werewolf/luxury werewolf/synthwave werewolf

  • Robocop

  • Menorah-Bot

  • Kamen Rider, SHOCKER villain in a cheap rubber monster outfit

  • Beer can Moses

  • Big Bird

  • Doctor Dreidel

  • Dreidel dominos

  • Greg Valentine

  • The lovely and dangerous Miss Elizabeth

  • Mercenaries

  • Spirit guardian

  • Sausage bird (an entity, a dish, and a vehicle. Badass)

  • Bobbing for sausage in a vat/tub of brown gravy (side note: the sausages are sausage bird and not pork. Perfectly fine for Hanukkah cards)

  • Kicking/punching/slapping (75% of the time, this entails a backwards hand or foot by design. I apologize that it is not 100%)

  • The NWA world’s heavyweight champion of Christmas

  • Swank swinger

  • Business dogs

  • Etc.

What is this?

A nice LoRA if you ask me. You choose 3 or 4 random words from the above and get something hilarious. You can also do really nice stuff!

Otherwise, there is a ridiculous amount of flexibility in this LoRA, but it is also a little unstable by design- it gives it more of an SDXL feel with unpredictability with Flux quality- awesome. Originally, this was going to be only slapstick style holiday portraits, I trained it, noticed some things worth expanding this into something more, and retrained a few times until I was satisfied. I leaned into some more ridiculous things from generating images with the first couple iterations, updated the dataset with superior high-res images from that, etc.