Like the model ❤️
CHECK NEW VERSION DETAILS!
BUZZ HELPS TO TRAIN MORE MODELS
Welcome to my first LoRA by FluxDev!
Here's the difference between using the LoRA and not using it... Does it convince you?
Honestly, I've been looking for LoRAs with vintage/aesthetic poses for Pinterest or Tumblr... Since I couldn't find one, I decided to create it myself. I hope you like it...
I trained it with only 2500 steps, and so far, the testing results are promising.
Description
1st version of Aesthetic Poses for FluxDev
FAQ
Comments (7)
Amazing! is perfect for dark enviroments with artificial lightning!!
While i like the poses, it unfortunately ruins the likeness of any face lora's I use :(
ooouch, this is becouse is trained with asian models... maybe you can generate the pose that you want and apply controlNet (openPose or something), im just training another version right now that maybe is more flexible, i have to finish the training and test how it works.... stay tunned!
i upload a new model, try with this one... but i tried to use one of my loras with the version 1.0 and it works good for my lora at least.
@MarTz_Arts Thanks for the response. That update was insanely fast! I'll test it out and let you know. im basically using this model in conjuction with celeb loras like crash's aerith lora in case you wanna know for testing purposes
@Spooder yeah , i have my own lora based on sabrina carpenter. I will upload images generated using that lora cambined with this one--- you will see the prompting, or maybe the loras that you are using are overcooked.
Maybe you should use 0.8 or 0.6 lora weight, and dont use all the triggers, use aesthetic and artificial lightning.
Perhaps you could only train certain layers. As I know 7 and 20 are for likeliness of the face. If you leave them out, the lora should interfere much less.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.