The truth is that when I made the dataset, I didn't do it as I usually do, and the result doesn't convince me, so I will surely make a version 2, with a dataset that I like more.
trigger words: todoroki shoto,1boy,male focus,split-color hair,white hair, red hair,short hair,scar,scar on face,burn scar,heterochromia,grey eyes,blue eyes,hair between eyes,
I upload my LoRA to civitai and pixai.art, so please don't upload to other sites.
I upload the LoRA for personal use, not for commercial purposes either the model or the images produced.
If you can give me 5 stars, I want to feel like an influencer for 5 minutes.
Description
the dataset was 50% screenshots of the anime and 50% fanarts, including so many fanarts, gives a result that is not my style, in version 2, I will change the dataset.
FAQ
Comments (4)
This looks amazing, would you please share how you get such great results? like how do you prepare dataset and which model you use for training
I only use the guide on how to make LoRa, which is here in civitai, and regarding the dataset I do it by selecting the images (especially the face of the character), I don't follow the method of just downloading images from gelboru by lots.
@Aki21 Thanks for sharing! do you tag images manually ?
@andinmaro146 No, I use the default tags, what I do manually is to edit the images in the dataset, cutting if there are other characters in the image.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.