Join our Discord community where you can generate videos, share ideas and request LoRAs, to push the bounds of the technology!
HUNYUAN v1.0 :
Trained on Hunyuan Video fp8 with 12 images of Mikasa in various episodes of Attack on Titan.
The trigger words are: M1K45a_4ck3rm4n woman, Mikasa Ackerman, Attack on Titan
Recommended Settings:
LoRA strength = 1.0
Embedded guidance scale = 7.0
Flow shift = 17.0
Prompt Recommendations:
Best to keep the prompt relatively simple, here are some examples that have worked well for me:
Example 1: 'M1K45a_4ck3rm4n woman, Mikasa Ackerman, Attack on Titan, walks through a destroyed city, flames reflecting in her cold, unwavering gaze.'
Example 2: 'M1K45a_4ck3rm4n woman, Mikasa Ackerman, Attack on Titan, brushes blood off her cheek, her expression unreadable against the flickering firelight.'
Example 3: 'M1K45a_4ck3rm4n woman, Mikasa Ackerman, Attack on Titan, sitting in a cozy café, sipping tea while reading a book, her red scarf neatly folded on the table.'
If it doesn't quite look like Mikasa, try also describing more features that she has, like 'black hair'.
Let me know if there are any questions, I'll be happy to help!
Description
Version 1.0
FAQ
Comments (5)
could you add a work flow? This looks awesome!! but all my results look like a modern Mikasa with swords and nothing AOT related. I'm also using one of the premade work flows and can't figure out where flow shift is. I used your recommended prompt.
Hey, of course I can, I could send you the JSON of the full workflow?
@Remade Thanks for the offer man, super cool of you. I decided I needed to figure it out myself like a big boy, I was using a workflow without the flux guidance node and when I added that and turned it to 7 like in you recommended it works now! you rock
@bman807 Really glad to hear that, props to you! Appreciate the kind words also :)
Awesome model,cheers,but i can't get the same results than you have.Could you share the JSON of the workflow? if you can't ,i can understand, thanks again ;)