For the Roman Gala: Don't Just Arrive. Ascend.
This season, the true influencer knows the toga is history. The future of neo-classical chic has a new name: the erotoga.
Forget shapeless drapery. The erotoga is pure, architectural elegance. Its stunning asymmetric cut sculpts your form, baring a shoulder with daring confidence. The elegant sweep of the sinus creates a powerful silhouette, while the glint of the signature fibula isn't just a detail—it's a statement.
While others are lost in cliché costumes, you'll command the room with a look that is both timeless and fiercely modern. It’s the perfect canvas for your content, a guaranteed scroll-stopper, and the undeniable it-piece that separates the follower from the leader.
This isn't a costume. It's a coronation. Show up in the erotoga.
How to Use This LoRA
Activation & Trigger Word:
The main trigger word is: erotoga
Additional trigger words for the details are: fibula, lacinia and sinus.
Controlling the Appearance:
The color of the erotoga can be controlled by simply stating it. The most common color would be white or off-white. You can try to specify a material as well, but the LoKR wasn't trained with that in mind.
The second most important aspect is the fibula, the buckle or clasp that holds the fabric together. It is recommended to state it in the prompt as well when it should be visible. The typical fibula would be a circular brass fibula.
The lacinia is the fabric that flows down the front from the fibula. (The other end of the fabric on the back is also called lacinia and you might try to address it in the prompt, but the control is weak). The intention of the lacinia is to cover the crotch so that the pussy is hidden. With your prompt you can try to control its position and interaction with it, e.g. to hold it in place.
The sinus is the piece of fabric that runs diagonal across the front. Usually there is no reason to prompt for its control.
Further hints:
This clothing LoKR wasn't trained with the intention to fix the shape of the nipples or pussy. Please use a LoRA that is specialized for those areas when necessary.
And although this LoRK is giving mostly fine results, it is known to make mistakes. But those can be easily fixed inpainting.
Training details
Krea 2:
When you want to use it with ComfyUI, please use https://github.com/Comfy-Org/ComfyUI/pull/16857 (and give that PR a 👍 so that it gets pulled quickly);
using it with diffusers should work out of the box.LoKR with factor=16 (Krea2Attention=16, Krea2SwiGLU=16), linear_dim=10000, linear_alpha=1, use_scalar and full_matrix as well as DoRA (dora_wd=true)
9804 steps (4902 real steps * batch size 2; gradient accumulation 2)
Trained with Krea 2 Raw, to be used with Krea 2 Turbo
training time: 7 hours on one 5090, including validation image generation
regularization images were used
masked training with the faces masked away
each image had multiple captions for versatility
Qwen Image:
LoKR with factor=32 (Attention=32, FeedForward=16), linear_dim=10000, linear_alpha=1, use_scalar and full_matrix
2240 steps (batch size 1, gradient accumulation 4)
training time: 5:49 hours on one L40S, including validation image generation
regularisation images were used
each image had 4 captions for versatility
Description
When you want to use it with ComfyUI, please use https://github.com/Comfy-Org/ComfyUI/pull/16857 (and give that PR a 👍 so that it gets pulled quickly)
FAQ
Comments (1)
I think that's more "an almost decent sash" but it's fabulous! And quite in the spirit of many periods of Rome.







