Posted on June 30, 2020 | by Julian Eckerle
Thanks to digital image editors, vivid visual representations of products have long been an essential part of the online shopping experience in Fashion.
The process of creating these e-commerce ready product images hasn’t changed much in the last ten years and, to this day, involves a lot of highly repetitive and non-creative manual image editing steps.
One of the most time-consuming of these tasks is background removal.
Even the most highly skilled image editors spend hours removing photo studio backgrounds from raw fashion product images daily. The manual techniques used in standard photo-editing tools are varied; the unnecessary effort being put into the task itself is universal.
It’s time for a paradigm shift.
Why automate background removal work?
Fashion product images require attention to detail. Thus, concerns to deliver consistent output while not missing deadlines are present in every e-commerce image editing operation. Background removal is the most error-prone and time-consuming task in the process to get a raw product image from the initial photo shooting to e-commerce ready.
An automated service like autoRetouch ensures precision, quality, and consistency while at the same time delivering results up to 10x faster – twenty-four hours a day, seven days a week.
How does the Remove Background component work in autoRetouch?
Background removal is the fundamental first step to edit Fashion product images and is applied differently depending on the kind of output image you want to produce. Therefore, autoRetouch currently offers you three background removal methods in the Remove Background component, which are relevant for different kinds of worklfows:
Select clothing and model
Here, our artificial intelligence identifies a piece of clothing and the human model wearing it through a process called semantic segmentation. Once this main subject is identified, it automatically removes everything else, not identified as the main subject from the image, e.g., a photo studio background environment. An example workflow with this setting can be seen here.
Select this option if you want to create the hollow-man, neck-join, or ghost mannequin effect. Here, the machine learning model selects only the piece of clothing in an image and removes the rest, e.g. mannequin, human arms, photo studio background, etc. An example workflow with this setting can be seen in detail here.
Remove plain background
This option works best for already pre-processed images with a uniform, plain background. It is recommended for use if you want to repurpose product images for marketplaces, by, e.g., changing the resolution, or product positioning. An example workflow with this setting can be seen in detail here.
The ‘Smooth Edges’ toggle included in the ‘Remove Background’ component settings allows you to choose whether you want the system to smoothen the edges on the main subject in the image or to receive a binary cutout version. If hair or any other fine details are an important part of the image, make sure this setting is turned on.
What makes the autoRetouch ‘Remove Background’ component special?
Our service is designed around the concept of workflows. In a workflow, you chain up components, each representing an automated version of various otherwise manual retouching steps.
That means our Remove Background component, while already powerful by itself, usually is just one part of an automated image processing workflow for you.
In a workflow, you can also incorporate and automate additional image editing work, such as skin editing, ground shadows, etc. sequentially and, as such, migrate your image processing operation to end-to-end automation.
The biggest impact when moving from manual image editing to automated image processing with autoRetouch is the amount of time you save: think seconds, not minutes per image.
This becomes especially powerful when you also leverage autoRetouch’s bulk-processing capabilities (currently up to 100 images per workflow at once) in order to retouch many images at once.
Retaining creative control and getting started with automated background removal
Our machine learning models are continuously optimized, and we aim to provide an as perfect as possible background removal service for every Fashion product image.
Currently, the system performs best with apparel/clothing images shot in a photo studio environment and we’re working on supporting many more image types going forward.
It’s very important for us to make sure you are able to work easily with autoRetouch and that you retain full creative control of your images. Hence, we make it easy to export files from autoRetouch in PSD format.
In this way, you always retain full creative control and can apply any finishing touches for an image yourself, if needed, while at the same time profiting from the massive time and cost benefits of using our service.
In addition, you’re able to easily integrate autoRetouch’s automated processing into your own way of working with manual photo editing tools by leveraging the export capabilities.
Today’s online fashion retail environment is all about user experience. Better images convert more customers and ensure a positive brand experience.
Optimization starts with the willingness to try something new. So, start your automation journey with autoRetouch workflows today by trying the service for free with up to 10 images.
If you want to review a few workflow examples, feel free to take a look here.
Whether you use autoRetouch for background removal only workflows, or for fully end-to-end workflows including other steps like skin retouch, we’re keen to hear your feedback.