Edit: NEW VERSION RELEASED! Please check
https://civarchive.com/models/4405/millie-bobby-brown-20
This is Millie Bobby Brown embed trained in SD 1.5, for 500 Steps using Textual Inversion at a learning rate of 0.005:300,0.0005:500, Batch size of 8, which is an effective 4K Steps. The Dataset contains 13 images. Embed is 1 vector.
Type the name of the embed in prompt to use in your gens, its "MilBob1V" by default but you can rename it to whatever you please.
Will work well with any merge with SD 1.5 in it, and works best with the SD Vae.
I also have a Ko-fi for tips and commissions
https://ko-fi.com/seedspiller
Description
Retrained this version with 92 images, trained to 5K steps and a learning rate of 0.005:1000, 0.00005:5000. Embed has 3 Vector per token.
Type the name of the embed in prompt to use in your gens, its "MilBob3V" by default but you can rename it to whatever you please.
Works best with the "vae-ft-mse-840000-ema-pruned" VAE File.
I also have a Ko-fi for tips and commissions
https://ko-fi.com/seedspiller