This LORA was trained to render Gallade, #0475, the Blade Pokemon.
Description
Version 1.0
FAQ
Comments (2)
Very nice lora! It does basically everything a Gallade lora should do. It's very easy to get rid of the donut hips by putting "wide hips" in the negative prompt. It also handles different colour schemes pretty well (I love my pink Gallades!) but isn't trained to make shiny Gallade and there's no good way to prompt a different-coloured chest horn.
Some quirks I've found: if you wanna get rid of Gallade's green shirt, the "white body" tag doesn't always entirely work, and even with "green body" in the negative prompt, Gallade often ends up with this weird green shrug-type clothing. Also, the green shirt makes it very difficult to generate Gallade with certain types of clothing like leotards and one-piece swimsuits.
Also, the chest horn sometimes doesn't generate, which is difficult to fix. If someone is still working on this lora, I'd suggest adding "chest horn" (or whatever it's called) as a trigger word. Also, it seems like the rear chest horn is usually way bigger than the front.
And when generating multiple characters at once, this lora often puts a head crest on the wrong character and chest horns on both characters. But that's easy enough to fix with inpainting.
Hello and thank you for your review! Yes, I can understand that Gallade's green portions are difficult to remove, but this would be because the green-white dual coloration isn't clothing, but Gallade's body, unlike Gardevoir where the dress and green upper body segment can be interpreted as a type of organic clothing. It is difficult for the A.I. to interpret the green segments on Gallade as anything else because of Gallade's arm-blades, which are a critical part of Gallade's physiology. As for the chest horn not generating, this sounds like atypical behavior as I did not see this in testing, but may result from clothing that obstructs the chest horn or any other prompt that would cause a conflict in rendering the chest horn. And, well, the problems with generating multiple characters is usually a problem with Stable Diffusion in general, not necessarily any LORA, although I admit that this LORA was trained to generate a single subject.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.





