​

About the Provider

​

Model Quickstart

• A valid Qubrid API key
• Access to the Qubrid inference API
• Basic knowledge of making API requests in your preferred language

import requests
import base64

url = "https://platform.qubrid.com/api/v1/qubridai/image/edit"

headers = {
    "Authorization": "Bearer <QUBRID_API_KEY>"
}
payload = {
    "model": "Qwen/Qwen-Image-Edit",
    "prompt": "Add a rainbow over the mountains",
    "negative_prompt": "",
    "true_cfg_scale": 4,
    "num_inference_steps": 25,
    "seed": 50,
    "use_inpainting": "false"
}
files = {
    "image": open("image.png", "rb")
}
response = requests.post(
    url,
    headers=headers,
    data=payload,
    files=files
)
if response.status_code != 200:
    print(f"Error {response.status_code}: {response.text}")
    exit(1)
result = response.json()
image_base64 = result["edited_image"]
image_base64 = image_base64.split(",")[-1]
image_bytes = base64.b64decode(image_base64)
with open("edited_image.png", "wb") as f:
    f.write(image_bytes)
print("Image saved to edited_image.png")

​

API Generated Response

• Input Image

Edit Prompt: Change the text to “MIDNIGHT DRIVE”.

• Output Image

​

Qwen Image Edit

​

Model Overview

• It extends Qwen-Image’s text rendering capabilities to image editing tasks, enabling precise text edits directly within images.
• The model processes the input image through Qwen2.5-VL for visual semantic control and a VAE Encoder for visual appearance control, allowing both semantic-level and appearance-level edits in a single workflow.
• It supports text-guided image editing while preserving visual fidelity and layout.

​

Model at a Glance

​

When to use?

• Text-guided image editing with precise control
• Direct text addition, deletion, or modification in images while preserving original font, size, and style
• Both low-level appearance edits (with unchanged surrounding regions) and high-level semantic edits
• Consistent visual quality and layout preservation during edits

​

Supported Editing Capabilities

• Low-level visual appearance editing (adding, removing, or modifying elements while keeping other regions unchanged)
• High-level visual semantic editing (object rotation, style transfer, IP creation with semantic consistency)
• Precise bilingual text editing (Chinese and English)

​

Inference Parameters

​

Key Features

• Semantic and Appearance Editing: Supports both low-level appearance edits (with unchanged surrounding regions) and high-level semantic edits while maintaining semantic consistency.
• Precise Text Editing: Enables direct addition, deletion, and modification of Chinese and English text in images while preserving original font, size, and style.
• Dual Visual Control Pipeline: Uses Qwen2.5-VL for visual semantic control and a VAE Encoder for visual appearance control within the same editing process.
• Text-Guided Image Editing: Performs controlled edits through text instructions while preserving visual fidelity and layout.
• Strong Benchmark Performance: Achieves state-of-the-art performance on multiple public image editing benchmarks.

​

Summary

• It extends Qwen-Image’s text rendering strengths to precise image and text edits.
• The model supports both semantic-level and appearance-level modifications in a single workflow.
• It enables bilingual text editing directly inside images with high visual consistency.
• This makes it suitable for inference use cases requiring accurate, layout-preserving image edits.