Jon Devo column
Can artificial intelligence cull a set of images better than the photographer?
Most of my time behind the camera, looking for the best light and finding creative angles, brings me joy. I love taking pictures. I’m in my element when covering events, anticipating key moments and getting ‘the shot’. However, despite being someone who found their way into photography as a retoucher, I can no longer stand spending countless hours at a desk, selecting, processing and editing my images.
While some people revel in the post-production side of things, after two decades in image-making, I’ve found it’s become the most tedious element of my work. If only there was some way for the images to be sorted into ‘keep and delete’ automatically. It turns out there is – and the technology behind these new tools is becoming increasingly sophisticated at knowing what to cull and what to keep.
The influence of artificial intelligence continues to seep into every aspect of our lives, particularly as creatives. The past year has seen a rise in the number of software tools that can effectively auto-cull unwanted images as we import them – saving valuable time, which in turn could save us money.
Tools such as Kodak Professional Select and PostPro’s Wand plugin use cloud computing to sift through your images and mark the duds for deletion. Compressed-for-web versions of your shots are sent to a cloud-based computer, which analyse the technical quality of the images to detect overor underexposure. They also assess the white balance and sharpness of the image to help keep image sets consistent and eliminate any obviously out-of-focus images.
They are also able to scrutinise the aesthetic qualities of each image and choose between near-identical shots. For example, if your portrait set includes similar poses, with some frames ruined by blinking and halfsmiles, auto-culling software will favour the best image unaffected by closed eyes and uncommitted facial expressions.
In use, I’ve been impressed with just how accurate these tools are already – and over time, these tools will only become more adept at deciding which images are keepers and which ones need to go. Given the speed at which artificial intelligence is improving, it will soon be commonplace to have virtual assistants handling all basic image processing and image selection tasks.
I do have one reservation about conceding yet another area of image creation to AI, though. Picture desk assistants and editors are trained to assess which images work best, based on their experience and understanding of audiences. Photography apprentices are also an invaluable part of the professional imaging world, and it’s often one of the main ways in which new talent can get a foot in the door. Reliance on AI could eliminate a crucial role that assistants, apprentices and picture desk workers perform.
“I’m impressed with just how accurate these tools are already”