Generative AI & Creativity: new photography styles?


Can generative ai create a new photography style? By new, I mean a style that is not widely known or established, something new and unknown in the photography world. AI is already making us reconsider what we mean by the term creativity. Probably because we always naively assumed that creativity was a uniquely human trait. Generative AI is rewriting all the old rules.

I used generative ai (chatGPT-4 and Dall-E3) to create a series of photographs of a base theme “a couple sailing at sunset on a calm lake”, using a few different photographic techniques. Shown below, these are tilt-shift photography (creates a “toy town” visual effect), macro lens photography (emphasizes closeup textures like the ropes on the decks below), slow shutter technique (to blur parts of an image, in this case an interesting twist where the static cloud objects are blurred perpendicular to the object moving across the field of view), and an HDR (high dynamic range which results in a highly vibrant image) photograph. Take a look, and let’s continue.

If a human took any of these photos, we would congratulate them on being a creative person. When ai does it, does it still feel creative? Or do we seek troubled comfort telling ourselves it just copied a set of style rules that humans invented in the first place?

So I raised the bar higher for generative ai and asked “We created images earlier in several photographic styles (tile shift, slow shutter, hdr, and macro). Can you create a new camera photography technique, name it, explain what defines it and why it’s unique, and then create the images in that new photographic style?”

Certainly! Let’s invent a new photography technique and call it “SpectraSilhouette.”
SpectraSilhouette Photography Technique:
Unique Trait: This technique combines long exposure with a spectrum gradient filter to capture silhouettes against a dynamically colored background.
•   How It’s Done: The camera is set to a long exposure time while a filter that transitions through a spectrum of colors is placed in front of the lens. As the exposure continues, the filter shifts gradually, infusing the background with a seamless gradient of hues. Subjects are positioned so that they are backlit by a strong light source, creating distinct silhouettes.
•   Why It’s Unique: SpectraSilhouette creates images where the subject is in stark contrast, depicted in pure black, while the background is a smooth transition of colors resembling the spectral dispersion of light. No two images are the same due to the unpredictable nature of the color transition during the exposure.

Now, let’s create images using the new SpectraSilhouette technique:

I looked at the image, and admit it’s not like other photography techniques I’ve seen. But it also seemed like it was a mix of digital art – and less photography – was generative ai cheating? I wondered if it had lost track of the photography requirement. So I asked it to explain the steps needed to create this image photographically with a camera:

“How would you create that using an actual camera, filters, lenses, and any post-processing steps?”

To create the second image using actual camera equipment, one would employ a combination of techniques, including:
1.  Silhouette Capture: Position the camera so that the subject is backlit by the sunset, ensuring the camera settings expose for the sky. This would naturally turn the subjects into silhouettes.
2.  Time-lapse Technique: Take a series of photos at intervals to capture the changing colors of the sky as the sun sets.
3.  Color Filters: Use color gradient filters to accentuate or alter the hues present in the time-lapse photos, possibly stacking different filters for a unique effect.
4.  Long Exposure: For the water’s silky texture, use a slow shutter speed to smooth out the ripples and waves over time.
5.  Post-processing: Combine the silhouette and the best color gradients from the time-lapse in post-processing software, layering the images to create the final composite that showcases the transition of time and color.
6.  Selective Sharpening: Apply sharpening to the silhouettes to ensure they are well-defined against the soft, color-blended background.
This combination of techniques would allow a photographer to replicate the SpectraSilhouette style with real-world camera gear and techniques.

That’s remarkable in a couple ways. First it figured out how to create a relatively novel photography technique. And Second, even though it created the image digitally, it understood how the technique would be created physically. Third, it did both in about 10 seconds.

What can generative ai do to discover new techniques in business, strategy, services, manufacturing, product design, economics, and every other field in which we’re engaged?

Recent Posts