Este LoRA fue entrenado con un dataset de 40 imágenes.
Para las imágenes se uso el modelo AnyLoraCleanLinearMix-ClearVAE y también se utilizo Hires.fix con lo siguiente:
4x-AnimeSharp
Hire Step de 10
Denoising de 0.25
Hire Scale de 2
Para estos LoRAs es recomendable poner estos prompts en Negativos mas el peso que recomiendo para que al menos no aparezcan los marcos y demás en el 99% de las imágenes generadas:
character name, card (medium), star (symbol), sun symbol, card parody, copyright name
Peso recomendado: 0.7 - 0.5
Tambien puedes utilizar mis LoRAs en TensorArt
Description
FAQ
Comments (2)
i see that you are using my prompts, but, to be honest, you should use all of them and not just some of them..
your negative prompting is really shittty and everyone can see this on your uploads.
those uploads and those hands are just horrible
Lol, me había olvidada de este comentario.
But you were the one who recommended me to use those 3 negative prompts when you told me at the time (I didn't even go to see your profile after that), lol
I used all those prompts from the articles you wrote and they didn't help me much because the hands still came out defective.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.