Logo
LoginGet Started
Profile

fireredteam / firered-image-edit

An advanced image editing model that modifies images based on text prompts. It supports single-image edits and multi-image compositions for tasks like style transfer or virtual try-on.

0.0325 per megapixel of image
GitHub
License
Weights

Model Input

Input

The editing instruction describing what changes to make to the image. Supports both English and Chinese instructions.

The URLs of the images to edit. Supports single image editing and multi-image references (e.g., virtual try-on, style transfer from reference).

The negative prompt for the generation.

Min: 2 - Max: 50

The number of inference steps to perform. More steps generally produce higher quality results.

Min: 1 - Max: 10

Classifier-free guidance scale. Higher values make the model follow the prompt more closely.

The same seed and the same prompt given to the same version of the model will output the same image every time.

Min: 1 - Max: 4

The number of images to generate.

The size of the generated image. If not provided, it uses the input image dimensions.

The format of the generated image.

Additional Settings

Customize your input with more control.

The acceleration level to use for inference speed optimization.

Safety checker can only be disabled on API call

You need to be logged in to run this model and view results.
Log in

Model Output

Output

Fill in the input form and click submit to see the output
Logs (1 lines)

Model Example Requests

Examples

00cqJToGCYrZvJXFvz3g99wAB4U9OzgEbTz4ApozMt

Model Details

Model Details

FireRed Image Edit is a state-of-the-art model that allows you to modify images using simple text instructions. This versatile tool can handle a wide range of editing tasks, from simple adjustments to complex compositions involving multiple images. By providing a descriptive prompt, you can guide the model to perform precise changes, making it an intuitive solution for creative image manipulation.

The model's core strength lies in its ability to understand and execute edits based on natural language. Whether you need to alter a background, change an object's color, or add new elements, a clear instruction is all it takes. Furthermore, FireRed Image Edit excels at tasks that require multiple image references. You can seamlessly combine subjects from one photo with the background of another, transfer the artistic style from a reference image, or even create virtual try-on scenarios by placing clothing items onto a model.

### Key Features - **Instruction-Based Editing:** Use text prompts in English or Chinese to describe your desired modifications. - **Multi-Image Referencing:** Combine elements or transfer styles from two or more source images. - **Fine-Tuning Controls:** Adjust parameters like `guidance_scale` to control how closely the model follows your prompt and `num_inference_steps` for image quality. - **Multiple Outputs:** Generate up to four variations of your edit in a single run to find the perfect result.

### Tips - For best results, write clear and specific prompts. Instead of "make it better," try "change the background to a sunny beach with blue skies." - When combining images, use high-quality sources for a more seamless blend. - Use the `seed` parameter to generate reproducible results if you need consistency across edits.

### Example Usage To run this model using the JavaScript client: ```javascript import { modelrunner } from "@modelrunner/client";

const result = await modelrunner.subscribe("fireredteam/firered-image-edit", { input: { prompt: "Make trees grow on the given book", image_urls: [ "https://example.url/example-image.png", "https://example.url/example-image2.jpg", ], num_inference_steps: 30, guidance_scale: 4 }, });

console.log(result); ```