) (((holding a chinese flag))) in a new.jpeg)
).jpeg)
)) (((as a toddler))) at soccer field with his mommie, iphone footage. 4k photo by david fincher.jpeg)
)). Compassionate,Empathetic,.jpeg)
)). Compassionate,Empathetic,.jpeg)
)), and mascara, (((at summer))). bloody nose,holding and looking at a ((meuse)) ((ext.jpeg)
)). Compassionate,Empathetic,.jpeg)
).jpeg)
)). Compassionate,Empathetic, Alex Proyas,Daring, Heroin, new scenes lift.jpeg)
)). Compassionate,Empathetic,.jpeg)
) while ordering coffee at Starbucks.jpeg)
Type | |
Stats | 185 |
Reviews | (3) |
Published | Apr 30, 2023 |
Base Model | |
Training | Steps: 5,000 |
Trigger Words | Tuckercarlson Person |
Hash | AutoV2 F91BBCD4D4 |
American Commentator/News anchor Tucker Carlson - High quality Dreambooth training using JP's DB repo.
Put "Tuckercarlson Person" in your prompt. Use prompt weighting if necessary for example "Tuckercarlson person ((wearing a santa hat))" would put more weight on the santa hat part in the event the model is too strong.
Yes, it's a full checkpoint as tagged. If you like lora stuff and feel that's on par w/ DB for likenesses or have limited resources, this download probably isn't for you.
For best results use Hires Fix. For ultimate results use in tandem w/ 30-60% the steps modified by Controlnet w/ Canny, Openpose, or shuffle. Controlnet.
SAMPLES WERE GENERATED W/ CONTROLNET (if you don't understand what that means you will probably be disappointed).
I find it fun to put actual movie still frames in the CN input and use it for the first half or the generation. If you have a quick system try turning on Generate forever w/ 2 or 3 selected for the batch. While it's running you can change the prompts and CN input/settings and A1111 will update and continue churning out generations.