Resource used: https://civarchive.com/models/22530
Your characters may have insufficient training data! Click here.
This is a requested character.
Weight: 1.0
Trigger + Appearance: narusegawa naru, antenna hair
Sample Outfit: yellow shirt, short sleeves, red skirt, pantyhose
Add brown hair and brown eyes if necessary.
The results of your generations are dependent on the additional LoRAs, weights, and models you use, so it may not work or come out as consistent as my sample images. Other reasons may include me using xformers, token merging, and a modified DPM++ 2M Karras sampler.
Description
Initial release.
FAQ
Comments (3)
Oh man! Classic. Natsukashii~
Man you're on fire today! Dropping so many loras at once!
how do you train so many loras? is there a way to automate this or something while you afk?
CORRECTION (5/14/23): I'm a Colab Pro user (basically paid access to better GPUs like A100), so when I exhaust all of my computing units—the thing you get with Pro or Pro+ and lose whenever a notebook is running, I am downgraded to a free user. To get the best out of this plan, I prepare the datasets for 20 LoRAs then upload them to Google Drive. With Pro, it takes on average under 10 minutes to train a character and 30ish minutes to train a style. As a free user, multiply those values by about 4.
Regardless of the plan, when a model is done baking, I prepare the sample images and then start training the next model. Once all models and images are ready to go, it's time to upload the models.