Component: Remove Background

The Remove Background component removes the background of any number of similar fashion or product images.

This article will dive into the options you have within this component and some things to keep in mind while using it.

Please note:

  • Remove Background is currently delivering the best results on fashion product images where a product is worn by a human model & with images where the clothing is presented on a mannequin or standalone.
  • PNG Input with already transparent background, currently result in a bad clipping of the object or model.

Examples:

Remove Background Settings

  • Method Selection: allows you to define the type of input image and expected output
  • Smooth Edges: smooth the edges of the remaining object
  • Vector Paths: choose to include vector paths for the outline of the remaining object

Method SelectioN

There are 5 different options for how to remove the background:

Select clothing & model

Select clothing & model removes the studio background entirely and captures the figure, in this case, both the human model as well as the clothing.

Select only clothing

Select clothing is trained to select only clothing, particularly on mannequins, but also on when worn or held by humans. It’s perfectly suited to help create ghost mannequin images.

Remove plain background

Remove plain background roughly removes a generic plain background where a model with a product, or just a product, is in the foreground. The background removal works best if the background is plain, preferably just a plain white.

You can use this setting in combination with the Set on Canvas component to quickly re-align already processed Fashion product imagery for example to achieve corporate identity consistency across your online shop or experiment with new background colors.

Select Object

This can be used for anything other than apparel and model shots.

Select Food

The select food algorithm is trained to pick up only food.

Smooth Edges

By selecting the smooth edges component setting, the system will automatically smoothen the visual transition between the foreground and background of a Fashion product image for you. If you need a binary background removal, simply toggle the switch off.

Vector Paths

If you would like to include a vector path of the outline of the object, expand the settings box labelled “vector paths” and select “create vector path.”

Please note, as the default export type is PNG, you must include the “export” component to download vector paths. There are 3 file types that support them: JPEG, TIFF and PSD.

Zapier integration

Zapier lets you connect autoRetouch with Cloud Spaces like Google Drive, Dropbox, etc., so you can automate your work and have more time for what matters most—no code required.

To set up the automation you need a autoRetouch and a Zapier Account.

Note: You will need to pay to use Zapier. You can find the pricing here: Zapier Pricing
To process one image it takes 2 Tasks inside Zapier.

Step by Step Guide

This Guide will show you how to set up a “Send new images to an autoRetouch Workflow” Template provided on Zapier.com. To receive the finished result back to your cloud storage you also need to set up the related “Upload newly retouched image” Template. The process is similar.

1. Select the autoRetouch/Zapier Template

Choose one of our pre-configured Zapier Templates to start setting up the automation.

Note: To send a Image from a Cloud Storage to autoRetouch and receive the retouched image back you need to set up both Zapier Templates.

2. Connect your Cloud Storage and create folder

Create an Input and Output Folder inside your Cloud Storage (OneDrive, Google Drive, etc.). The automation will be triggered as soon as a new image is uploaded to a specified folder in the cloud. The second Zapier Template will upload the result image to a specified folder.

Connect the Zapier automation to the according folder.
(Note: If asked to include file object in Zapier, select YES)

3. Connect to autoRetouch and select Organization

Connect your autoRetouch Account to Zapier and select the Organization and the workflow you want to automate.

Note: The workflow still has to be set up inside our autoRetouch Web App.
Map Cloud Storage and autoRetouch fields

4. Map Cloud Storage and autoRetouch fields

Use the following File Name and File Objects inside Zapier:

Cloud StorageFile NameFile
OneDrive“File Name”“File”
Dropbox“File Name”, “File Ext”“File”
Box“Name”“File”
Google Drive“Title”“File”

5. Set Zapier automation active

Set the Zapier automation on active. The Upload from the Cloud Storage to autoRetouch will automatically be triggered if a new image is added to the specified folder.

FAQ

My Google Drive Images are not automatically uploading!
If the modification date of the new file is older than the last newest item, the Google Drive app will not recognise it as new. This is a known issue of the Google Drive app and haven’t been fixed yet.

Can I upload a file to autoRetouch and download it back to my Cloud Storage in one Zapier?
No. Due to the architecture of autoRetouch, it is not possible to create a Zapier automation that includes uploading to autoRetouch and downloading to your cloud storage. For this process you will need to use the two templates available here.

Can I trigger more then one workflow within one Zapier automation?

No. You will need to create separate workflows in autoRetouch for each variant of the image you wish to create, which you will need to trigger through multiple Zapier automations.

How much Tasks does one image cost me?
As you need two Zapier automations (One to send the image and one to receive it back to your system) you need two Zapier Tasks per Image.

Is there an alternative way to connect autoRetouch to my System?
Yes. You can always use our public API to connect your systems directly to autoRetouch.

Input & Output

autoRetouch supports a variety of file formats and sizes.

Since we are focused on eCommerce Images in particular there are some current limitations on the input and output format.

InpuT

Supported Input Format: JPEG, JPG, PNG, TIFF, TIF, WEBP

Maximal File Dimensions: 10k x 10k pixels

Notice:

  1. Everything above 4096 x 4096 pixels will be scaled down to fit into a 4096 x 4096 pixels frame
  2. When trying to insert a PNG with a transparent background the remove background component is currently not working

Output

Supported Export Format: JPEG, JPG, PNG, TIFF, TIF, PSD (For Details see: Export)

Maximal File Dimensions: 4096 x 4096 pixels

If you are interested in using autoRetouch on a high volume but need a format we are currently not supporting feel free to reach out!

API FAQ

Q: Do you have a public API?

A: Yes. It is documented here: docs.api.autoretouch.com

Q: Do I need to contact you for getting an access token/API key?

A: No. Your normal autoRetouch account is enough. Just use the documented device authorization flow to get everything you need.

Q: What ClientID do I need for API access?

A: V8EkfbxtBi93cAySTVWAecEum4d6pt4J

Yes, it is the public documented one (docs.api.autoretouch.com). Yes it is the same for all API accessing clients (expect some internal autoRetouch ones). No there is no customer specific one. The name is miss leading, but cannot be changed due to OAuth/authorization backend requirements.

PSD Export Options

Note: If you export multiple images as PSD files, please allow for more processing/downloading time. These files are much larger than a .jpg or .png.

How to export as PSD file

  1. Go to the “Workflow Builder”
  2. Drag & drop the export component into the workflow. The default output is .png, to change this go to the export settings and select “file type” PSD.
  3. Note: the export component must be the last one in the workflow.
  4. Select the options you want to be included in your PSD file (explained below):
  • Include guides: these help to align layers with the Set on Canvas padding and borders
  • Include layer mask: choose this option when you want to make further changes to images with areas of “fuzziness,” for example, on hair. Only works with the “remove background” component.
  • Include input image: will help you to have all information in one place, and no dependencies to other files. It also lets you compare autoRetouch Results with the input image.
  • Include set on canvas layers: if you want to make changes to your custom background.
  • Include vector paths: only works with the “remove background” component. Choose this option when you want to make further changes to images with clear and defined borders.
YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Layer structure

When multiple components are included in the workflow, the export will contain different kinds of corresponding layers:

  • Remove bg component can export 2 layers: cut out image, and cut out image processed
  • Set on canvas component exports 3 layers: image, downsized image, and background layer
  • All other components export a single layer
  • The lowest layer is the input image in the original size that autoRetouch processed the image in.
  • Canvas size is determined by the size of the biggest layer, to enable you to always see all of your content while editing.
YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Layer Masks

Layer masks will apply to the output of the remove background component. They are optimal for making changes to the border of fuzzy areas of an image, like hair.

This is how you will see your exported results in photoshop:

To make changes to the alpha mask in Photoshop:

  1. Zoom in on the object
  2. Hold “option” on mac / ”alt” on windows and click on the mask
  3. Select brush (shortcut: B)
  4. Choose a color that matches the opacity of the paint you want to use.
    Note: An alpha mask uses various shades of black, white, and grey to show the opacity of the mask.
  5. Paint over the part you want to be removed with color matching the opacity.
  6. Click back onto the layer to see the final result.
YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Vector Paths

Vector paths are good for making adjustments to images with a defined edge, for example when no hair is involved.

Update: the option to include vector paths is now located within the component creating the path (remove background and skin retouch component).

How to make changes to paths in Photoshop:

  1. Select the input layer and disable the cutout layer
  2. Go to paths
  3. You can then see the vector path
  4. Select the “path selection” tool
  5. Select “path”
  6. With layer and path selected, go to layer → vector masks → current path
  7. The path will then be applied to the layer
  8. Switch to “direct selection” tool
  9. Can then select path points and change the edge (you can even select single path points
YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Layer Guides

If after downloading your image you want to make changes and place the image back onto the “set on canvas” layer later, you can export guides to help with that.

  1. Select “include guides” in the export component and process the image(s).
  2. Open the processed image in Photoshop
  3. You will see layer guides around the “set on canvas” layer as well as guides that correspond to padding defined in autoRetouch

How to make changes with layer guides in Photoshop:

  1. Select mask
  2. With the brush tool, make desired changes to the mask
  3. Duplicate the layer
  4. Right click mask → apply layer mask
  5. Move to top and disable the rest
  6. Use the layer guides as orientation to move image and snap to the layer guides
  7. Select “crop image tool” and scale image down to set on canvas guides
  8. Create a new layer
  9. Select the “fill tool”
  10. Grab the background color and fill the background
YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Set on Canvas Layers

Use this feature if you want to have a specific background color or a custom background image, and you need to make changes to something after processing:

  1. Add the remove background, set on canvas, and export components to your workflow.
  2. Set on canvas settings → background → can add color or upload an image.
    Note: If you select a custom background image, the background image size will determine the canvas size for your image.
  3. Select “include set on canvas layers” in the export component

How to make changes to the set on canvas layers in Photoshop:

You will see 3 “set on canvas” layers:

  • The processed layer is the final result (object applied on the background image)
  • The foreground layer is the object (as scaled by set on canvas to corresponding size)
  • The background layer is the custom background image (size corresponding to the background image size)

To make changes, select the foreground or background layer, make the necessary changes, and enable the other one.

YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Exporting out of Photoshop

Scenario 1 – no set on canvas / face crop included (no resizing of image)

  1. Hide every layer except the top layer
  2. The canvas that defines your image size will correspond to the input and output dimensions that you want.

Scenario 2 – set on canvas / face crop component included

When either of these components are included, the image will need to be resized

  1. Select the “canvas/background“ layer
  2. Make sure top layer is visible
    Note: If you have defined your own background image which has some transparency, you could see through this and may have to hide the layers below it.
  3. Take the crop tool and crop the canvas layer to match the top layer size
    Note: Including guides in the export options in this step will insure the crop tool will snap to the guides, and make this step much faster.
  4. If the resized image is smaller than the input image, then you will need to crop the canvas before exporting. If it’s bigger, you don’t need to do anything.
  5. Export your image.

Ghost Mannequin Effect

What is it all about?

In the fashion world, a ghost mannequin (aka neck-joint, invisible mannequin, or hollow-man) is the name for an image editing technique that creates the illusion of a human body filling clothes for product images in e-commerce.

YouTube

By loading the video, you agree to YouTube’s privacy policy.
Learn more

Load video

Functionality:

We automatically create a finished product image for you based on two initial product photography shots. The product and the inlay shot.

Warping intensity

Shadow intensity

Price:
Each processed ghost mannequin image costs €0.50 and is charged against your credit balance.

Supported images

We currently support:

  • Simple Dresses
  • Skirts
  • Pants, Shorts & Bathing trunks
  • Tops, T-shirts & Pullovers
  • Jackets & Outerwear
  • Underwear & Bikini bottoms

We currently do not support:

  • Open jackets
  • Brassiere
  • Multiple items at once
  • Complex shapes

Ghost Mannequin Guidelines

For the recommended Photostudio set up and the expected input images please have a look at our Ghost Mannequin Guidelines.

Best Practice

For more best practices have a look at the Guidelines linked above.

Thinks to avoid

FAQs

Can I download the image without the watermark?
Yes. Click “process image”. The image will show in the result list below.
This action will charge your credit balance.

Can I still process other images?
Yes. You can process any other image (model photography, etc.) via your defined workflows.

Can I download a PSD file?
Yes, you can adjust your settings to choose between a PNG file or a layered PSD file.