☑️If you like my work, leave a review. I really appreciate them!✍️
Jlullaby [PonyXL] Style DoRA. I recommend using 1.0 Strength and using this with AutismMix_Confetti for best results.
Trigger Word: jluby
Please look at prompts used in example images for help! Read About This Version for more information.
I use additional tools/ extensions for my generations so my images may be different from yours. If you want to know what I use, please look at my guide.
My Socials | Support me on Patreon or Ko-fi (If you want)
Description
Initial Release
Use jluby when generating
This is based off their semi 3d/ realistic style. More maybe in the future...
FAQ
Comments (6)
I have heard this DoRA technique is actually better than LoRa? Is that actually true and what is it? Is it just a new training technique or maybe something else entirely?
It basically replaces speed for added detail when training. It is much better for styles and closer to being trained like a checkpoint. If you were to merge this with a checkpoint, the style would work better than a style for a lora unless it's really overtrained.
How does DoRA fair comared to LoRA? Especially if you are using regional prompt and the weight needs to be toned down a lot for any art style to work. I have a main art style and was thinking of using this as a 2nd, but wanted to ask. It would also fit in the lora files correct after installation?
A DoRA may overfit more than the regular lora. DoRAs are closer trained to a checkpoint compared to a lora. You may need to use a lower strength when using but it all depends on how the other style you are using is trained.
And yes, you put this into the regular spot that other loras go. If it does not work, you may need to update the program you are using to generate images as it may not support it.
@guy90 Continuing from this, aside from the similarities between the artist and what not, does it also improve general hands and feet, or is it the art style that is the main catch for this? Since some art styles may not work too well and mess up hands and stuff, I was just wondering if it just improves on everything that a LoRA would have.
@AkashaMoka Hands depend on the artist, the amount of hands in the dataset, and how under/ overtrained it is. If there is a low amount of hands in the dataset, it's somewhat hard to do hands. It also depends on how the artist draws hands and how consistent the style is. A DoRA just does a better job at retaining the style.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.