Ti depicting Melissa Manzoni. A pretty model with lovely face and entrancing eyes. There is an NSFW element in the training data, so a negative NSFW/nude prompt may be necessary for SFW generations.All sample images use either ADetailer (face), Highres. fix (x2) or a combination thereof; no other embeddings, LoRa or manual inpaints were used. Entire- or majority-body generations used NSFW/nudity negative prompts; the closeup/portrait images used no negative prompts (viewing image prompt metadata is all accurate for generation purposes). Update: see version information for v1.1 details.TI was trained on a dataset of 96 publicly-available images with a batch size of 12 and gradient accumulation of 8 steps per epoch for 370 epochs using the scheduler ADAMW with constant LR of 0.01. Image masking was leveraged to focus the training. Minor skin touchup was performed and a wrist tattoo removed from the training data.All models depicted in training data are over the age of eighteen years at the time of the creation of such depictions per 18 U.S.C. § 2257.Would love to see some creations using this embedding!

Melissa Manzoni

v1.1 - same dataset, batch size, gradient accumulation, etc., but lowered LR to 0.005; the training ran for 220 epochs in total. Generally-speaking, the outputs are very similar to v1.0, but is prone to less artifacting on freckles/blemishes and has better freckle consistency overall. I think the face shape is better in v1.0, so nearly didn't bother uploading this version.

Files with SHA256: fde289b98de067305d3c679d42afed94cc36967b30f4e082042995c9d9fed752