# MOSS Style v1: Photorealistic LoRA for MOSS-Image Base

A photorealistic style LoRA for MOSS-Image Base. This is v1.

## What it does

It handles almost any kind of generation:

- Everyday life / daily-life scenes

- General portraits and lifestyle shots

- Strong selfie-style images (this is where it's best)

- Average-level NSFW: it can do it, but the NSFW side is not very strong in v1

## Usage

- Trigger word: [none]

- Strength: 0.8-1.2 (start at 1.0)

- Base model: MOSS-Image Base (INT8 ConvRot build works too)

## Recommended settings

- Steps: 25

- CFG: 3

- Sampler: euler

- Scheduler: simple

- Negative prompt: empty (Base default)

- Resolution: 1024x1024, 720x1280 (9:16), 832x1216, etc. (multiples of 16)

The workflow is embedded in the example images. Drag a PNG into ComfyUI to load it.

## Notes

- This is v1. If you'd like a v2, tell me your feedback and wishes in the comments.

- If you notice any quality issues, post your results on my Discord: https://discord.gg/Up2fukT83

- Trained on: curated photo set