The very first Machine Learning tools to Photo, Pencil Tool improvements to Designer, and Multi-Page Spreads and other advanced page management updates to Publisher are included in the Affinity 2.6 update.

The updates announced by Serif, developers of Affinity, are free to owners of version V2 of the software and you should automatically be prompted to update to the latest version of the software when opening your app. The update is for the three apps, Affinity Designer, Affinity Photo and Affinity Publisher, so once you’ve updated one you should do the same for the other two, if you’ve the whole package which continues to be available for one low bundle price, USD$164.99 – and single payment – that includes the three apps on all operating systems, including iPad.

After joining Canva, the company made four pledges to the community and one of those pledges was that they would provide Affinity to eligible schools and nonprofits for free. That is true since September last year, when it was announced that eligible nonprofits and K-12 (primary and secondary) teachers and schools can download the entire Affinity suite with a full license at no cost. This includes full access to Affinity Photo, Designer and Publisher on Mac, Windows and iPad for as long as you have a valid Canva For Education or Canva for Nonprofits account.

Machine Learning that works only “on device”

Affinity 2.6 has landed and this version brings some very exciting updates and improvements to the software, including Serif’s very first Machine Learning features, which are present in the Photo app and come with an important note from the company: the two new features which use Machine Learning models (Object Selection Tool and Select Subject) are optional and require downloading of the relevant models for them to work. These are installed as pre-trained models and they do not use any of your own data for further training. Furthermore, these operations all work ‘on device’ meaning none of your data leaves your device at any time.

Affinity apps use data models to perform automatic object and subject selections on a pixel layer or placed image and the Machine Learning models function as ready to install sets of data that you can optionally download to allow Affinity to automate object and main subject selections in your compositions.

The company also says that “if you’re worried that you will run out of storage on your device, you can uninstall the models at any time to restore space on your hard drive. Ensure to reinstall them if you wish to use the ML features again.”

Affinity’s Machine Learning models are designed to enhance workflow efficiency by harnessing the power of machine learning. Affinity says that by automating object and subject selection tasks, users can benefit from the following:

• Speeding up repetitive tasks, allowing users to focus on creative aspects more.
• Reducing errors when manually selecting composition objects and their parts.
• Setting automatic subject selections as a Macro to jump to editing part of users’ workflow faster.

According to Serif, there is no additional cost involved to install the models. Your Affinity apps come with the models ready to install if you wish to use automatic object and subject selection features. In terms of system requirements for Machine Learning, these models and linked functionality is only available on Apple Silicon macOS devices running the most up to date operating system (macOS Ventura 13.0 and newer) as well as on Windows x64 and Arm64 hardware running Windows 10 or 11.

The new Machine Learning features

With Object Selection you can speed up your editing process by using this new Machine Learning (ML) tool that automatically selects objects of your choosing in an image. This is in line with Affinity’s ambition to use ML for the benefit of faster workflow. As noted above – and it is good to state it again – models are installed as pre-trained models and do not use any of your own data for further training.

The Object Selection Tool can be accessed via the Tools panel. It uses a Segmentation Machine Learning model to identify objects on a layer. Initially, while hovering over an area, the tool will indicate the inference processing by changing its icon temporarily. Once ready, the hatched lines will appear over the object to indicate what will be selected when clicked. Every newly selected area is treated as a new object selection.

The new Select Subject option is a Machine Learning feature that analyses your image and selects what it considers to be the main subject of the scene. The operation can be recorded as a Macro to aid setting up foreground/background retouching workflows. The Select Subject feature can be accessed via the top menu’s Select entry. It uses a Saliency Machine Learning model to identify the most prominent subject on a layer.

Several usability improvements

The Select Subject feature does not have any customisable options. However, if you wish to add or subtract from the selection, use Affinity’s raster selection tools with an appropriate mode set. You can also refine subject selections, mask, and work on subject selection areas as you normally would when using those tools.

Users of the Affinity Publisher app will be glad to know that spreads can now contain more than two pages, making it easy to create trifold, gatefold, accordion-fold and other page arrangements. One note of caution: the implementation of this feature involved a rewrite of much of the existing page, spread and master code. As such, 2.6.0 cannot support backward compatibility to 2.5 for page operations. Affinity explains this in detail in its support forums.

Affinity Publisher has a series of other improvements regarding page management. Affinity Designer’s Pencil Tool has also changed and now includes several new options on its context toolbar. A number of usability improvements have also been made to Affinity Photo, something especially beneficial for users familiar with other image editors. The update also includes a few other minor improvements that are mentioned in the website page about version 2.6.

https://affinity.serif.com/en-us/whats-new/