# Refmod character likeness — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/c2ffe215ef8689be

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-10-24 16:47 ingested 2026-10-25 02:38

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

## retained text



```
Can folks share best practices they have run into for consistently getting near perfect character likeness using refmods? I have experimented a few times and have had some success but it has felt a bit inconsistent. Feels like it’s pretty dependent on getting the perfect set of input photos (and maybe even photo order?) while creating the refmod, and prompting ofc feels very important as well. Welcome any tips here and any set ups/ workflows folks are using to get high quality / near perfect likeness results via refmods. Also my experience is that refmods is the better/easier way to get near perfect character likeness compared to classic ref2va but wanted to confirm that is other folks’ experience as well. Thanks all!
```



document fetched 2026-10-25 00:29 · canonical https://reddit.com/r/StableDiffusion/comments/1xediro/refmod_character_likeness
