Skip to main content

Remove watermarks from an image

Watermark Remover automatically detects and removes watermarks from images. Watermarks can be solid (or) see-through text and logos.

This transformation supports png, jpeg, jpg, webp, cr2, nef, rw2, dng, orf, raw, heic, heif, avif, tiff and tif type of files.

User can specify upto 5 regions in form of boxes in the image where the watermark can be removed.


By default, Watermark Remover will only remove transparent watermarks. To remove non-transparent texts or logos, use rem_text and rem_logo parameters.

Params

Remove text (rem_text)

Remove non-transparent textual information from an image.

Remove non-transparent logos from an image.


info

The rem_text and rem_logo parameters are not mandatory fields. If both are false, only the watermark will be removed.

Boxes (box1, box2, box3, box4, box5)

Select regions for texts or logos removal by adding the starting position and dimensions in the box fields. You can select a maximum of 5 different boxes at once for an image. By utilizing the 5 boxes, you can achieve greater precision in selecting the portion of the image to remove the watermarks. The removal will only occur in the selected area.

The box parameters define the starting position and dimensions of your selection. The input value is of the form x-axis_y-axis_width_height and these values are specified in percentage (%). These percentage values are concerning the image's dimension.

As illustrated in below Image 1, the value of the box is 0_0_50_100 and point A is the starting point of the image, from where the x-axis and y-axis begins, as shown with the arrow pointed. So the box will start covering the image's 50% width and 100% height from where the arrow points, as illustrated in Image 1 with the blue box.

Image 1

In the below scenario, as illustrated in Image 2, the value of box is 20_0_60_100 and A is the starting point of the image, from where the x-axis and y-axis begin, as shown in Image 2. Here the x-axis is 20% and the y-axis is 0%, so the box will start covering its 60% width and 100% height from where the arrow points, as illustrated in Image 2 with the blue box.

Image 2

If two or more boxes have overlapping values, then that will not have a negative impact on the outcome, as shown in the below diagram with various examples.

Watermark Removal Boxes

Multiple Boxes

In the below three scenarios, two boxes are used to show their effect when removing the watermark.

Fully Overlapping

In this scenario, two boxes are used. The value of box1 is 0_0_100_50 which has a blue border and covers the upper half of the image, while the value of box2 is 0_0_100_20 which has a red border and covers the top 20%. The two boxes overlap, as illustrated in the below diagram.

Fully overlapping boxes
Transformed Imagewm.remove(rem_text:true,box1:0_0_100_50,box2:0_0_100_20)

Partially Overlapping

In this scenario, two boxes are used. The value of box1 is 0_0_50_100 which has a blue border and covers the left half of the image, while the value of box2 is 0_0_100_20 which has a red border and covers the upper 20%. The two boxes partially overlap, as illustrated in the below diagram.

Partially overlapping boxes
Transformed Imagewm.remove(rem_text:true,box1:0_0_50_100,box2:0_0_100_20)

Not Overlapping

In this scenario, two boxes are used. The value of box1 is 0_0_100_40 which has a blue border and covers the upper 40% of the image, while the value of box2 is 0_50_100_40 which has a red border and covers 40% of the image in height starting from the center of the image. The two boxes do not overlap, as illustrated in the below diagram.

Not overlapping boxes
Transformed Imagewm.remove(rem_text:true,box1:0_0_100_40,box2:0_50_100_40)

Transform and enhance your images using our powerful AI technology. Organize your images in more efficient manner and our extensible APIs enables seamless integration with your system unleashing the power of our platform. Join the large community of users who use PixelBin to transform their image libraries and achieve excellent performance

Is this page useful?