# Topaz Image Enhance

Topaz Image Enhance applies industry-standard upscaling and image enhancement to a single input image using Topaz models. It sends the image to the Topaz API, processes it with the selected model, and returns the enhanced result. You can choose from four models: Reimagine, Bloom 2, Wonder 3.5, and Wonder 4.

## Inputs

### Common Inputs

| Parameter | Description | Data Type | Required | Range |
|-----------|-------------|-----------|----------|-------|
| `image` | The input image to enhance. Only one input image is supported. | IMAGE | Yes | Single image |
| `model` | The Topaz enhancement model to use. The selected model determines which model-specific settings appear. | DYNAMIC_COMBO | Yes | `"Reimagine"`<br>`"Bloom 2"`<br>`"Wonder 3.5"`<br>`"Wonder 4"` |
| `output_width` | Zero value means to calculate automatically (usually it will be original size or scaled proportionally to `output_height` if specified). Wonder 3.5 and Wonder 4 support upscale factors from 1x to 6x, Bloom 2 from 1x to 12x. Bloom 2, Wonder 3.5 and Wonder 4 always keep the input aspect ratio: if both sizes are set and their ratio differs, the larger scale wins, so the output is bigger than the requested size and the price is higher than it would be for the requested size. (default: 0) | INT | No | 0 to 32000 |
| `output_height` | Zero value means to output in the same height as original or scaled proportionally to `output_width` if specified. Wonder 3.5 and Wonder 4 support upscale factors from 1x to 6x, Bloom 2 from 1x to 12x. Bloom 2, Wonder 3.5 and Wonder 4 always keep the input aspect ratio: if both sizes are set and their ratio differs, the larger scale wins, so the output is bigger than the requested size and the price is higher than it would be for the requested size. (default: 0) | INT | No | 0 to 32000 |

### Reimagine Inputs

These settings apply when `model` is set to `"Reimagine"`.

| Parameter | Description | Data Type | Required | Range |
|-----------|-------------|-----------|----------|-------|
| `prompt` | Optional text prompt for creative upscaling guidance. (default: "") | STRING | Yes | Any text |
| `creativity` | Creativity level for the enhancement. (default: 3) | INT | Yes | 1 to 9 |
| `subject_detection` | Subject detection mode (advanced). | COMBO | Yes | `"All"`<br>`"Foreground"`<br>`"Background"` |
| `face_enhancement` | Enhance faces (if present) during processing. (default: True) | BOOLEAN | Yes | true<br>false |
| `face_enhancement_creativity` | Set the creativity level for face enhancement. (default: 0.0) | FLOAT | Yes | 0.0 to 1.0 |
| `face_enhancement_strength` | Controls how sharp enhanced faces are relative to the background. (default: 1.0) | FLOAT | Yes | 0.0 to 1.0 |
| `face_preservation` | Preserve subjects' facial identity. (default: True) | BOOLEAN | Yes | true<br>false |
| `color_preservation` | Preserve the original colors. (default: True) | BOOLEAN | Yes | true<br>false |
| `crop_to_fill` | By default, the image is letterboxed when the output aspect ratio differs. Enable to crop the image to fill the output dimensions. (default: False) | BOOLEAN | Yes | true<br>false |

### Bloom 2 Inputs

These settings apply when `model` is set to `"Bloom 2"`.

| Parameter | Description | Data Type | Required | Range |
|-----------|-------------|-----------|----------|-------|
| `prompt` | Optional text prompt for generation. Leave empty to auto-generate a prompt from the input image. (default: "") | STRING | Yes | Any text |
| `creativity` | 1 is restrained enhancement, 9 is pronounced reinterpretation with newly generated detail. (default: 3) | INT | Yes | 1 to 9 |
| `seed` | Seed for reproducible generation. (default: 2) | INT | Yes | 1 to 2000 |
| `color_preservation` | Preserve the original colors. (default: True) | BOOLEAN | Yes | true<br>false |
| `grain` | Add grain to the output image. (default: False) | BOOLEAN | Yes | true<br>false |
| `grain_model` | Grain model to use. Is ignored if grain is disabled. | COMBO | Yes | `"silver"`<br>`"gaussian"`<br>`"grey"` |
| `grain_strength` | Strength of the grain effect. Is ignored if grain is disabled. (default: 0.5) | FLOAT | Yes | 0.0 to 1.0 |
| `grain_size` | Size of the grain particles. Is ignored if grain is disabled. (default: 1.0) | FLOAT | Yes | 1.0 to 5.0 |
| `grain_density` | Intensity of the grain effect. Is ignored if grain is disabled. (default: 0.5) | FLOAT | Yes | 0.0 to 1.0 |

### Wonder 3.5 Inputs

These settings apply when `model` is set to `"Wonder 3.5"`.

| Parameter | Description | Data Type | Required | Range |
|-----------|-------------|-----------|----------|-------|
| `enhancement_strength` | Enhancement level for varying input conditions. (default: "high") | COMBO | Yes | `"low"`<br>`"medium"`<br>`"high"` |
| `grain` | Add grain to the output image. (default: False) | BOOLEAN | Yes | true<br>false |
| `grain_model` | Grain model to use. Is ignored if grain is disabled. | COMBO | Yes | `"silver"`<br>`"gaussian"`<br>`"grey"` |
| `grain_strength` | Strength of the grain effect. Is ignored if grain is disabled. (default: 0.5) | FLOAT | Yes | 0.0 to 1.0 |
| `grain_size` | Size of the grain particles. Is ignored if grain is disabled. (default: 1.0) | FLOAT | Yes | 1.0 to 5.0 |
| `grain_density` | Intensity of the grain effect. Is ignored if grain is disabled. (default: 0.5) | FLOAT | Yes | 0.0 to 1.0 |

**Note:** Only one input image is supported; the node raises an error if the input batch contains more than one image. Grain settings (`grain_model`, `grain_strength`, `grain_size`, `grain_density`) are ignored unless `grain` is enabled. For Bloom 2, leaving `prompt` empty automatically generates a prompt from the input image. Wonder 4 exposes only the common inputs. The output is capped at 179 megapixels and at 32000 pixels per side.

## Outputs

| Output Name | Description | Data Type |
|-------------|-------------|-----------|
| `IMAGE` | The enhanced and upscaled image returned by the Topaz API. | IMAGE |

> This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! [Edit on GitHub](https://github.com/Comfy-Org/embedded-docs/blob/main/comfyui_embedded_docs/docs/TopazImageEnhanceV2/en.md)

---
**Source fingerprint (SHA-256):** `dd109d31ea4ab7928058414c3a62ae3829f879354ebdce329b599aacb9804d1b`
