Resizing

The Claid API allows you to set the final image size for different workflows you may have. By default, API will perform all calculations respecting the original image aspect ratio. Meaning, the image can be cropped but it won't be stretched or squeezed.

When you need a size bigger than the original image, you can use upscale parameter to increase the final image resolution

Width and Height

Specifies the size of the output image. The width and height parameter supports dynamic sizing and is based on pixels or percent values.

The default behavior of API resizing

• If width and height are not specified the image size will not be changed.

• If you use default fit option(crop) and specify only one side of the image(width or height), the unspecified side remains unchanged and you get a cropped image.

• If you use a fit option other than the default crop and specify only one side of the image(width or height), another side will be defined as auto and the final image will be resized keeping the aspect ratio.

• When width and height are specified the resulting image will be resized until it fully covers the specified dimensions and gets cropped.

The default behavior can be changed by the fit property.

Acceptable values:

Value	Type	Description
"auto"	string	The dimension will be calculated automatically to save the aspect ratio of the input image. Both width and height cannot be set to auto. When using auto, specify at least one of the dimensions in pixels or percentages.
1000	integer	width or height in pixels.
"200%"	string	Percentage as a string. E.g. "200%" means that the dimension will double relative to the input image.

Examples

You can read more about "fit": "bounds", which is used in the following examples in Fit section

Empty

JSON Body

"operations": {
    "resizing": {
        "width": 4000,
        "fit": "bounds"
    }
}

Auto

Pixels

Percentage

Fit

The fit parameter controls how the image will be constrained within the provided size, relative to the width and height. Check the image examples below to see how fit works.

Possible properties:

bounds Preserves the aspect ratio. Resize the image to fit entirely within the specified region, making one dimension smaller if needed.

cover Preserving aspect ratio. Resize the image to entirely cover the specified region, making one dimension larger if needed.

canvas Input image will be placed on the canvas established by width and height. The aspect ratio of the input image won’t change. Extra space will be filled with the background color.

crop Scales an image until it fully covers the specified dimensions, the rest gets cropped. Fit crop support center and smart modes for cropping. See the Smart Crop section of this page to learn more.

Outpainting

outpaint Generates a coherent background for an image to fit specified dimensions, scaling an image until the larger side reaches the edge of canvas established by width and height first, so generation happens on either left-right or top-bottom sides symmetrically.

If you are looking for outpainting all four directions simultaneously - see Zoom Out (Outpaint on all sides).

The feathering parameter of the fit object for outpaint type allows you to adjust the smoothness of the transition from the original image content to the generated area. The feathering value is the ratio of the transition area to the total width or height of the input image, depending on what is being resized. In most cases there will be two generated areas on the output image, so "50%" is a limit as it will turn the entire input image into a transition area.

"operations": {
    "resizing": {
        "width": "150%",
        "height": "100%",
        "fit": {
            "type": "outpaint",
            "feathering": "20%"
        }
    }
}

outpaint fit type currently has size limitations: the target image size must not exceed 16.78 Mp (e.g., 4096 x 4096 pixels, 5461 x 3072 pixels). Our team is working on increasing max size

Zoom Out (Outpaint on all sides)

Expands an image in all four directions simultaneously—top, bottom, left, and right. The effect is like zooming out of the original shot, creating a consistent, larger background around the entire image.

To use this, set the fit type to outpaint and provide values for the outpaint_by parameter. You can specify how much to expand in each direction either by percentage of the input image size or by fixed pixel values.

Value	Type	Description
"10%", or "5% 15%"	string	Fills the space relative to the total image size. Can be specified: A single value for all sides. Two values for the horizontal and vertical sides. ❗ The value cannot exceed 50%.
"30px", or "100px 150px",	string	Fills the space with the exact number of pixels. Can be specified: As a single value for all sides. Two values for the horizontal and vertical sides. ❗ The value cannot exceed 50% of the image's width or height, respectively.

The values in outpaint_by are added to each side of the image, not split — this applies to both pixels and percentages.

So "100px 50px" means 100 pixels on the left and right, 50 on the top and bottom.

"10% 5%" means 10% of the image width added to each side, and 5% of the height to the top and bottom.

A 1000×1000 image with "10% 5%" would become 1200×1100.

The width and height parameters still define the final output size. When using outpaint_by, these values are applied after the outpainted area has been generated.

If you only want to extend the image by a fixed amount in all directions — without resizing the final image — you should either set both width and height to "auto", or omit width and height entirely.

This means you can:

1. Control how much padding is added around the image using outpaint_by.

2. Scale the outpainted result using width and height.

Ensure that the restorations.upscale property is set to an upscaling method of your choice. See Upscale from more info.

"operations": {
    "resizing": {
        "width": "200%",
        "height": "auto",
        "fit": {
            "type": "outpaint",
            "feathering": "20%",
            "outpaint_by": "20% 10%"
        }
    }
}

outpaint fit type currently has size limitations: the target image size must not exceed 16.78 Mp (e.g., 4096 x 4096 pixels, 5461 x 3072 pixels). Our team is working on increasing max size

Possible values:

Value	Type	Description
crop	string or object	DEFAULT. Scales an image until it fully covers the specified width and height, the rest gets cropped. Content-aware cropping.
bounds	string or object	Scales an image until the larger side reaches the edge of canvas established by width and height.
cover	string or object	Scales an image until the smaller side reaches the edge of canvas established by width and height.
canvas	string or object	Puts input image on the canvas established by width and height. Aspect ratio and size of the input image won't change. Extra space will be filled with the specified background color.
outpaint	string or object	Scales an image until the larger side reaches the edge of canvas established by width and height, then fills the extra space with a generated coherent background.

Examples

Crop

JSON Body

"operations": {
    "resizing": {
        "width": 1440,
        "height": 1440,
	"fit": "crop"
    }
}

Bounds

Cover

Canvas

Outpaint

Zoom Out

Smart Crop

Fit crop supports center and smart modes for cropping.

center is a basic cropping mode that crops images from their center without considering their content.

smart is a content-aware cropping mode that detects a main object in a photo and uses it as the center point for cropping.

crop: center compared to crop: smart

Smart Crop ensures output photos are perfectly framed and focused on the main object.

Center

JSON Body

"operations": {
    "resizing": {
        "width": 1440,
        "height": 1440,
	"fit": { 
	    "crop": "center" 
	}
    }
}

Smart