> For the complete documentation index, see [llms.txt](https://docs.pletor.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.pletor.ai/models/image-models/krea-3.md).

# Krea 3

Krea's newest image model, built for fast, instruction-following generation and editing that keeps products, characters and text consistent.

## Overview

Where [Krea 2](https://docs.pletor.ai/models/image-models/krea-2.md) is built for art direction, Krea 3 is built for following instructions: describe a new image or an edit in plain language and it changes what you asked while keeping products, logos, characters and text intact. A prompt expander rewrites your prompt and, with aspect ratio on Auto, picks the framing. Up to 2K; edits combine up to 8 references.

### Strengths for marketers

* **Product consistency**: a product in a new scene, shape, label and logo kept.
* **Consistent characters** across poses, locations and lighting.
* **Precise edits**: an object, material, color or piece of text, nothing else.
* **Multi-reference edits**: up to 8 references (product, model, setting) in one image.
* **Sketch to finish**, with instructions for materials and light.

### Ideal use cases

* Product photography across lifestyle scenes.
* One campaign face across a set.
* Iterative retouching.
* Concept to final.

### Weaknesses

* **2K cap**: pair with an upscaler for print.
* **No style references or moodboards**: use Krea 2 to lock a brand look.
* **Auto ignores the resolution setting**: the expander frames and renders at about 2048×2048.

***

## How to use effectively

Describe the change directly ("Change the mug to red, keeping everything else the same"). Identify each reference by content ("the sneaker from the second image on the person in the first"). Name what must stay intact. Pick 2K for small text or finals. Raise **Reasoning** to High for complex edits (offered only with an image connected).
