---
tags:
- ' Realistic'
- stable-diffusion
- ' Semi-Realistic'
- Photorealistic
- SDXL
---
This model is a LORA for improving humanoid women bodies in SDXL. 

Dattaset:

It was learned on ~500 perfect pictures (without any unperfections) with repetition 10 * batch 5 * 4...8 epoh, and ~500...1500 very good pictures (with most features very good) for diversity with lower attention (repetition 1..4).
All womens in range 18-26, most was 20yo.

Trained on base of SDXL1.0fix but work best on sdxlUnstableDiffusers_v5, and new versions on v8.

Best practives and findings:
* Dataset shoud be perfect in all aspects and each feature. Dont think SD has smart AI from what you can prompt good parts and do negative bad, no, but each bad pixel seems do model worse.
* Big diversity lead to problems, i redraw 200+ images by hand to make some consistency of nipples and this improve result a lot.
* Only simple body positions, without many body parts intersections or perspective distortions. No fisheye, macro, 

Failed attempts:
* low network alpha lead to very slow training, after 3k-5k steps quality almost no improve, probably need to search best value in range 0.5-0.9 of dim (same as dim mean turn it off).
* scale_weight_norms=5 seems lead to overtrain, 2 is better
* Network dim lower then 64 seems not enought to this dataset
* LoKR recommended Multires noise discount 0.8 is not working from start

Version history:

all standart LoRA, prodigy preset, scale buckets, no regularization images

v0.25 no spec activation. Partially fixed problematic features. 1750 images, network 128/96, scale_weight_norms=2

v0.2 no spec activation. may or not use 1girl. 1300 images. learned with prodigy and adafactor preset, network 128/16. Same problematic lips and nipples due dataset diversity and probably overtraining on some pictures.

v0.1 activation: g01tt. 1100 images, network 32/16, but has some overtraining issues but undertrained in same time. Problematic nipples.