# Has anyone tried the Krea2TextEncoder for reference-based gens? (some recent results) — retained record

> Retained reddit record from reddit:StableDiffusion.

Canonical: https://highsignal.app/data/records/e8273a9a9cace621

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


source: [reddit](https://highsignal.app/data/reddit) (reddit:StableDiffusion)  published 2026-11-17 12:15 ingested 2026-11-18 02:39

[original source ↗](https://www.reddit.com/r/StableDiffusion/comments/1xyixro/has_anyone_tried_the_krea2textencoder_for/)

## retained text



```
Been playing around with ComfyUI-Krea2TextEncoder lately (paired with Krea 2 Turbo) and honestly really liking this approach for reference-based generations.

What I like most about this method:

 **- Zero sampling overhead:** The vision/text embeddings are extracted upfront into conditioning before sampling even starts, so it doesn’t slow down the diffusion loop at all.

**- Almost no prompt wrestling:** You don’t have to spend time describing lighting, outfits, or poses in a wall of text - the encoder reads the reference and handles most of the scene for you.

Have any of you tried this encoder yet, and what kind of results did you get?
```



document fetched 2026-11-18 00:32 · canonical https://reddit.com/r/StableDiffusion/comments/1xyixro/has_anyone_tried_the_krea2textencoder_for
