I have trained two versions of this LoRA with the same dataset, but with two different trainers: Simple Tuner and Kohya_ss. I prefer the second version --- but try it yourself :) Have fun!
First of all: HOLY COW! FLUX is sooo good. Thanks for this model and thanks to the AI community for updating things like ForgeUI: Luv ya all. <3
Anyway! :D
This LoRAs were/are made for human skin, good hands and good eyes.
Here I share the training steps: 1300, trained with Simple Tuner.
Here I share the the first epoch, trained with - you already guessed it^^: Kohya_ss.
If you need more flexibility, please join my PATREON page, where you get all training steps and the final LoRAs and how to use it:
https://www.patreon.com/polyhedron_ai
Description
V.10
FAQ
Comments (5)
Ah great you ported this. I used it a lot on SDXL
Yes. Still working. :D Thank you.
This has happened a few times while using this model, notice a bar below the photo: https://imgur.com/bWTAcMD
I have never seen this problem before. My guess: The training resolution is/was 1024 x 1024; for this I cropped or resized some images and inserted a black, white or sepia background to fill the gaps - this could not be recognized by Simple Tuner; I will retrain this LoRA with Kohya_SS soon; maybe your problem will disappear then. Stay tuned!
It's online. <3
Details
Files
Available On (4 platforms)
Same model published on other platforms. May have additional downloads or version variants.