# Moss Image Showcase – Mostly Photoreal + Some Art — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/0fb7d26b52558f49

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-10-23 09:49 ingested 2026-10-24 02:38

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1xd9o13/moss_image_showcase_mostly_photoreal_some_art/)

## retained text



```
Each and every image is generated using MOSS IMAGE base model no lora or fine-tune.

Settings at cfg 4 and steps mostly stayed at 39 very few images were made at 42 or 50 steps,each and every image is generated at fix seed=11932(for reproducibility purposes),samplers used ranged between euler,sa\_solver and sa\_solver\_pece and scheduler was fixed at sgm\_uniform,shift at 5 and SLG=off.resolution ranges from 1.33MP to approx 2MP in some images.

Images took longer time around 2.5-4 min cause I have a rtx3060

why not use turbo lora or sft?some may ask ,

The reason is simple turbo lora are still being trained and with few testings on them I came to a conclusion it may need some time as there were too many issues such as blur and out of focus and many other ,i didnt use sft or its turbo loras cause sft has vertical grid(which is solved by community btw) and other issue was artificial looks and ai feel due to may be supervised training of base model done on somewhat or more synthetic dataset....it's debatable some may prefer sft cause more better prompt adherence and less chaos compared to base.

I will soon share prompts as I manage to sort them out.

Here are my previous posts on moss image if anyone wants to check them out it was mostly focused on art and styles 

[MOSSzzART](https://www.reddit.com/r/StableDiffusion/comments/1x9dtuy/mosszzart/)

[MOSS-IMAGE showcase](https://www.reddit.com/r/StableDiffusion/comments/1x8awgl/mossimage_showcase/)


EDIT:Here are the prompts and settings one thing is default that I didn't mention in the pastebin is that scheduler is always sgm_uniform and cfg is 4 always

[click here for prompts and settings ](https://pastebin.com/imfn5FZU)
```



document fetched 2026-10-24 00:32 · canonical https://reddit.com/r/StableDiffusion/comments/1xd9o13/moss_image_showcase_mostly_photoreal_some_art
