Qwen-Image InstantX Inpainting ControlNet
Professional inpainting and image editing with Qwen-Image InstantX ControlNet. Supports object replacement, text modification, background changes, and outpainting.
简介
The Qwen-Image InstantX Inpainting ControlNet workflow is a powerful tool for professional image editing, specifically designed for tasks such as object replacement, text modification, background changes, and outpainting. At its core, the workflow utilizes the Qwen-Image model, which is renowned for its high-quality inpainting capabilities. The workflow integrates several nodes, including KSampler, CLIPTextEncode, and ControlNetInpaintingAliMamaApply, to facilitate detailed and precise image manipulations.
Technically, the workflow begins with uploading the necessary models and the image to be edited. Users can then define the areas for inpainting using the MaskEditor, which is accessed via the LoadImage node. The workflow employs the VAELoader and UNETLoader nodes to encode and decode images, ensuring that the edits blend seamlessly with the original image. The ControlNetLoader and ControlNetInpaintingAliMamaApply nodes are pivotal in applying the inpainting effects, allowing for intricate modifications while maintaining the image's overall coherence.
This workflow is particularly useful for creative professionals seeking to enhance or alter images with precision and ease.
你需要提供
- 正在读取工作流
在 ComfyUI 中运行。 此工作流无法在本页面运行。你可以在 Comfy Cloud 中打开（已预先加载所需内容），或下载后在自己的 ComfyUI 中使用。
在 Cloud 中试用查看工作流内部
节点图的只读预览。在 Comfy Cloud 中打开或下载后即可编辑节点。
Loading the graph
Open full-size workflow preview- 作者
- ComfyUI
需要自定义节点
- 正在读取工作流
通过代码使用此工作流
尽可能根据工作流图和输入自动生成代码，否则显示已保存的示例。运行代码可能产生费用，请先核对凭据和端点。
What you need
- Runs on
- Comfy Cloud
- 密钥
- COMFY_API_KEY
需要付费的 Cloud 方案和额度。
# Python 3.10+: pip install comfy-sdk==0.4.0
import os
from comfy_sdk import Comfy
api_key = os.environ.get("COMFY_API_KEY", "").strip()
if not api_key: raise RuntimeError("Set COMFY_API_KEY")
os.environ.setdefault("COMFY_BASE_URL", "https://cloud.comfy.org")
client = Comfy(api_key=api_key)
workflow = client.workflows.from_json({
"3": {
"class_type": "KSampler",
"inputs": {
"cfg": 2.5,
"denoise": 1,
"latent_image": [
"122",
0
],
"model": [
"66",
0
],
"negative": [
"108",
1
],
"positive": [
"108",
0
],
"sampler_name": "euler",
"scheduler": "simple",
"seed": 134554158057228,
"steps": 20
},
"_meta": {
"title": "KSampler"
}
},
"6": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"38",
0
],
"text": "The Queen, on a throne, surrounded by Knights, HD, Realistic, Octane Render, Unreal engine"
},
"_meta": {
"title": "CLIP Text Encode (Positive Prompt)"
}
},
"7": {
"class_type": "CLIPTextEncode",
"inputs": {
"clip": [
"38",
0
],
"text": " "
},
"_meta": {
"title": "CLIP Text Encode (Negative Prompt)"
}
},
"8": {
"class_type": "VAEDecode",
"inputs": {
"samples": [
"3",
0
],
"vae": [
"39",
0
]
},
"_meta": {
"title": "VAE Decode"
}
},
"37": {
"class_type": "UNETLoader",
"inputs": {
"unet_name": "qwen_image_fp8_e4m3fn.safetensors",
"weight_dtype": "default"
},
"_meta": {
"title": "Load Diffusion Model"
}
},
"38": {
"class_type": "CLIPLoader",
"inputs": {
"clip_name": "qwen_2.5_vl_7b_fp8_scaled.safetensors",
"device": "default",
"type": "qwen_image"
},
"_meta": {
"title": "Load CLIP"
}
},
"39": {
"class_type": "VAELoader",
"inputs": {
"vae_name": "qwen_image_vae.safetensors"
},
"_meta": {
"title": "Load VAE"
}
},
"60": {
"class_type": "SaveImage",
"inputs": {
"filename_prefix": "ComfyUI",
"images": [
"8",
0
]
},
"_meta": {
"title": "Save Image"
}
},
"66": {
"class_type": "ModelSamplingAuraFlow",
"inputs": {
"model": [
"37",
0
],
"sampling": "flow",
"shift": 3.1000000000000005
},
"_meta": {
"title": "ModelSamplingAuraFlow"
}
},
"71": {
"class_type": "LoadImage",
"inputs": {
"image": "image_qwen_image_instantx_inpainting_controlnet_input_image.png"
},
"_meta": {
"title": "Load Image"
}
},
"76": {
"class_type": "VAEEncode",
"inputs": {
"pixels": [
"172",
0
],
"vae": [
"39",
0
]
},
"_meta": {
"title": "VAE Encode"
}
},
"84": {
"class_type": "ControlNetLoader",
"inputs": {
"control_net_name": "Qwen-Image-InstantX-ControlNet-Inpainting.safetensors"
},
"_meta": {
"title": "Load ControlNet Model"
}
},
"108": {
"class_type": "ControlNetInpaintingAliMamaApply",
"inputs": {
"control_net": [
"84",
0
],
"end_percent": 1,
"image": [
"172",
0
],
"mask": [
"121:253",
0
],
"negative": [
"7",
0
],
"positive": [
"6",
0
],
"start_percent": 0,
"strength": 1,
"vae": [
"39",
0
]
},
"_meta": {
"title": "Apply ControlNet Inpainting (AliMama)"
}
},
"122": {
"class_type": "SetLatentNoiseMask",
"inputs": {
"mask": [
"121:253",
0
],
"samples": [
"76",
0
]
},
"_meta": {
"title": "Set Latent Noise Mask"
}
},
"124": {
"class_type": "MaskPreview",
"inputs": {
"mask": [
"121:253",
0
]
},
"_meta": {
"title": "Preview Mask"
}
},
"126": {
"class_type": "ImageCompositeMasked",
"inputs": {
"destination": [
"172",
0
],
"mask": [
"121:253",
0
],
"resize_source": False,
"source": [
"8",
0
],
"x": 0,
"y": 0
},
"_meta": {
"title": "Image Composite Masked"
}
},
"163": {
"class_type": "SaveImage",
"inputs": {
"filename_prefix": "ComfyUI",
"images": [
"126",
0
]
},
"_meta": {
"title": "Save Image"
}
},
"172": {
"class_type": "ImageScaleToMaxDimension",
"inputs": {
"image": [
"71",
0
],
"largest_size": 1536,
"upscale_method": "area"
},
"_meta": {
"title": "Scale Image to Max Dimension"
}
},
"121:199": {
"class_type": "GrowMask",
"inputs": {
"expand": 0,
"mask": [
"71",
1
],
"tapered_corners": True
},
"_meta": {
"title": "Grow Mask"
}
},
"121:251": {
"class_type": "MaskToImage",
"inputs": {
"mask": [
"121:199",
0
]
},
"_meta": {
"title": "Convert Mask to Image"
}
},
"121:252": {
"class_type": "ImageBlur",
"inputs": {
"blur_radius": 31,
"image": [
"121:251",
0
],
"sigma": 1
},
"_meta": {
"title": "Blur Image"
}
},
"121:253": {
"class_type": "ImageToMask",
"inputs": {
"channel": "red",
"image": [
"121:252",
0
]
},
"_meta": {
"title": "Convert Image to Mask"
}
}
})
# run() also passes api_key to the partner nodes in this workflow.
job = client.run(workflow, api_key=api_key)常见问答
Consider increasing the shift value in the workflow settings to reduce blurriness or darkness in the output.
Decrease the shift value to enhance detail, and ensure that your input image is not too large to avoid processing issues.
While the official number of steps is 50, using around 10 steps often yields satisfactory results without compromising speed.
It's recommended to use images smaller than 3000 pixels in dimension. Use the ImageScaleToMaxDimension node to resize larger images.