Stable Diffusion Upscale

Upscale the initial image to given shape. You can choose to restore the face within the image by turn up the visibility of gfpgan and codeformer (with upscaling_resize_ratio to 1)

Try it in the Widget Center

Click this url to try this widget and copy the Pro Config template.

Usage

Scale up the original image to the specified shape. You can opt to restore the face within the image by increasing the visibility of GFPGAN and Codeformer.

Input Parameters

Name	Type	Description	Default	Required
image	string	The input image, can be a url or base64 sting
resize_mode	integer	Sets the resize mode: 0 to upscale by upscaling_resize amount, 1 to upscale up to upscaling_resize_h x upscaling_resize_w.	0
gfpgan_visibility	number	Sets the visibility of GFPGAN, values should be between 0 and 1. Use for face restoring	0
codeformer_visibility	number	Sets the visibility of CodeFormer, values should be between 0 and 1. Use for face restoring	0
codeformer_weight	number	Sets the weight of CodeFormer, values should be between 0 and 1. Use for face restoring	0
upscaling_resize_ratio	number	By how much to upscale the image, only used when resize_mode=0.	2
upscaling_resize_w	integer	Target width for the upscaler to hit. Only used when resize_mode=1.	512
upscaling_resize_h	integer	Target height for the upscaler to hit. Only used when resize_mode=1.	512
upscaler_1	string	The name of the main upscaler to use	ESRGAN_4x
upscaler_2	string	The name of the secondary upscaler to use. Set to 'None' for quick upscale. The result will be a fusion of two model, decided by `extras_upscaler_2_visibility`.	None
extras_upscaler_2_visibility	number	Sets the visibility of secondary upscaler, values should be between 0 and 1.	0
upscale_first	boolean	Should the upscaler run before restoring faces?	False

Output Parameters

Name	Type	Description	File Type
url	string	The url of generated image, stored in the cloud. Only temporarily effective, will be cleared in a few hours.	image

Output Example

success

{
  "url": "https://image.myshell.ai/image/chat/embed_obj/40295/20240423/29381f8ee23144a69e88a64fa8232fca.jpg"
}

fail

throw a Http error.

Detailed Guidelines

• Upscale (SuperResolution)

  • Main parameter:

    • image : The input image, can be a url or base64 sting. Any size

    • resize_mode: Sets the resize mode: 0 to upscale by upscaling_resize amount, 1 to upscale up to upscaling_resize_h x upscaling_resize_w.

    • upscaling_resize_ratio : By how much to upscale the image, only used when resize_mode=0.

    • upscaling_resize_w & upscaling_resize_h: Target width/height for the upscaler to hit. Only used when resize_mode=1. Change the ratio of with/height will casuing a center crop of generated images

    • upscaler_1 : The name of the main upscaler to use, choose ["None", "Lanczos", "Nearest", "ESRGAN_4x", "LDSR", "R-ESRGAN 4x+", "R-ESRGAN 4x+ Anime6B", "ScuNET GAN", "ScuNET PSNR", "SwinIR 4x"]

    • upscaler_2: The name of the secondary upscaler to use. Set to 'None' to disable. The result will be a fusion of two model, decided by extras_upscaler_2_visibility

    • extras_upscaler_2_visibility: Sets the visibility of secondary upscaler, values should be between 0 and 1. The result will be final = (1-α) image_a + α image_b

  • Other functional parameter:

    • gfpgan_visibility : Sets the visibility of GFPGAN, values should be between 0 and 1. Use for face restoring. 0 for disabled. The result will be a fusion addition

    • codeformer_visibility && codeformer_weight : Sets the visibility and weight of CodeFormer, values should be between 0 and 1. Use for face restoring. The result will be a fusion addition

    • upscale_first : Should the upscaler run before restoring faces? default to False