> ## Documentation Index
> Fetch the complete documentation index at: https://docs.orisu.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Image models

> The image model families Orisu supports for generation, editing, and specialized image tasks.

Orisu's image nodes run on a set of model families. You pick the model in a node's config panel; this page lists what each family is for so you can choose. For what individual settings do (aspect ratio, resolution, guidance, seed, and so on), see [Model settings](/models/settings).

## Generation models

These families create images from a text prompt. Use them in the [Generate image](/nodes/ai-image/generate-image) node.

| Family          | Provider          | Good for                                                                                                |
| --------------- | ----------------- | ------------------------------------------------------------------------------------------------------- |
| Flux            | Black Forest Labs | General-purpose generation across speed and quality tiers; reference-guided and subject-consistent work |
| GPT Image       | OpenAI            | Prompt-faithful images with readable text and transparent backgrounds                                   |
| Nano Banana     | Google            | Fast iteration, multi-reference edits, and high-resolution (up to 4K) output                            |
| Seedream        | ByteDance         | Photoreal results and detailed scenes                                                                   |
| Imagen          | Google            | Photographic quality with fast and ultra tiers                                                          |
| Ideogram        | Ideogram          | Typography and text rendering, plus character-consistent generation                                     |
| Recraft         | Recraft           | Brand and design assets, illustration, and a built-in vector output path                                |
| Grok Imagine    | xAI               | Expressive, stylized images                                                                             |
| Higgsfield Soul | Higgsfield        | Stylized, aesthetic-driven portraits and scenes                                                         |
| Reve            | Reve              | Prompt-accurate general generation                                                                      |

### Variants

Each family exposes one or more variants you select per node. Higher tiers usually trade speed for quality and cost more credits.

<Columns cols={2}>
  <Card title="Flux">
    Flux 2 Pro, Flux 2 Max, Flux 2 Flex, Flux 2 Turbo, Flux 2 Klein, Flux Kontext Pro, Flux Kontext Max, Flux Krea, Flux LoRA.
  </Card>

  <Card title="GPT Image">
    GPT Image 1, GPT Image 1.5, GPT Image 2.
  </Card>

  <Card title="Nano Banana">
    Nano Banana, Nano Banana Pro, Nano Banana 2, Nano Banana 2 Lite.
  </Card>

  <Card title="Seedream">
    Seedream 4, Seedream 4.5, Seedream 5 Lite.
  </Card>

  <Card title="Imagen">
    Imagen4, Imagen4 Fast, Imagen4 Ultra, Imagen3.
  </Card>

  <Card title="Ideogram">
    Ideogram v3, Ideogram Character, Ideogram Reframe.
  </Card>

  <Card title="Recraft">
    Recraft 4, Recraft 4 Pro.
  </Card>

  <Card title="Single-variant families">
    Grok Imagine (xAI), Higgsfield Soul (Higgsfield), Reve (Reve).
  </Card>
</Columns>

<Tip>
  Ideogram Character and Nano Banana support subject references, so you can keep a character or product consistent across generations. Connect reference images in the [Generate image](/nodes/ai-image/generate-image) node.
</Tip>

## Editing models

The [Edit image](/nodes/ai-image/edit-image) node modifies an existing image from text instructions. Selectable editing models:

| Family       | Variants with editing                                  |
| ------------ | ------------------------------------------------------ |
| Flux Kontext | Flux Kontext Pro, Flux Kontext Max, Flux Krea          |
| GPT Image    | GPT Image 1, GPT Image 2                               |
| Nano Banana  | Nano Banana, Nano Banana Pro, Nano Banana 2            |
| Seedream     | Seedream 4, Seedream 4.5, Seedream 5 Lite, SeedEdit v3 |

## Background removal

The [Remove background](/nodes/ai-image/remove-background) node runs on **BirefNet** and returns a transparent PNG.

## Vectorize

The [Vectorize image](/nodes/ai-image/vectorize-image) node runs on **Recraft Vectorize** to convert a raster image into an SVG.

## Erase

Object and region erasing runs on **Finegrain Eraser**, which removes content from an image and fills the gap. Nano Banana also offers an erase path.

## Change camera angle

The [Change camera angle](/nodes/ai-image/change-camera-angle) node runs on **Qwen Camera Angle**, which reorients a scene using yaw, pitch, and zoom.

## Reframe and aspect ratio

The [Change aspect ratio](/nodes/ai-image/change-aspect-ratio) node reframes an image to a target ratio. It runs on a selectable model (Flux Kontext Pro, Nano Banana Pro, Nano Banana 2, or Ideogram Reframe), depending on whether you want the new area outpainted, cropped, or stretched.

<Note>
  Each model charges credits per run, and higher-quality tiers cost more. See [Model settings](/models/settings) for the options each model exposes and [Credits](/guides/credits) for how usage is billed.
</Note>