The Future of AI in Editing and Underwater Photography

Comprehensive Guide to Adobe Generative Fill (AI) for Underwater Photography.

AI-generated images, such as those displayed above, still have significant progress to make in accurately and realistically representing the underwater world, scuba diving, and marine life. Utilizing the beta version of Photoshop, the Generative Fill feature can produce images with only subtle indications of their artificial origin.

If you are a photographer utilizing Adobe Creative Suite products, such as Lightroom and Photoshop, it is likely that you are employing various AI-powered editing tools. The latest update, known as Generative Fill, is currently in beta testing within Photoshop and has the potential to revolutionize underwater photography.

Adobe has been integrating artificial intelligence into its editing workflow for several years. Each new release of Lightroom has enhanced the capabilities of its masking tools, particularly the automated masking features. For instance, the software can automatically select specific subjects, such as a diver or a manta ray, from an image. These masking techniques bear a resemblance to the portrait mode used by iPhone cameras and the background blur features employed in video conferencing applications.

 

Adobe Generative Fill offers the capability to incorporate objects, such as scuba divers, thereby enhancing the dimensionality of images.

Generative Fill elevates the editing process by leveraging artificial intelligence to enhance decision-making. This advanced tool can be utilized to add objects to an image, remove unwanted elements such as backscatter, generate backgrounds, and even extend the scene—essentially creating components of the image that previously did not exist.

 

How is Generative AI Used?

Using Adobe’s Generative Fill feature is quite straightforward. Begin by selecting the area of the image you wish to modify, which can range from a simple rectangular selection to a more complex layer mask that seamlessly blends with the composition. Once your selection is established, a toolbar will appear with a button labeled “Generate Fill.” By clicking this button, you can either enter a descriptive prompt or leave the field empty before selecting ‘Generate’ to view the results.

Let us examine three primary use cases for underwater photographers.

 

Adding an Object

Just pick the spot in the image where you want that object to pop up, toss in a snazzy description for the AI, and kick back while it whips up three dazzling variations! If those don’t tickle your fancy, no worries! Tweak your description and give it another whirl. Photoshop’s got your back, saving all those variations in the working layer, so you can always revisit a gem from earlier!

Removing a diver from a photograph using the Generative Fill tool in Adobe Photoshop is a straightforward process.

 

Removing an Object

Removing an object, such as a diver’s fin or a companion’s hand, follows the same methodology as adding an object, often functioning effectively without the need for a prompt.

Previously, backscatter could be eliminated either piece by piece or through the application of a layer mask in conjunction with the Dust & Scratches filter, which allows for the removal of large areas of backscatter simultaneously. This latter method is particularly effective when dealing with a simple water gradient. However, the smoothing effect of the filter becomes apparent when it intersects with intricate details, such as those found in coral reefs, subjects, or even background elements like water ripples.

 

In contrast, generative AI can efficiently eliminate backscatter, even in these intricate areas, with just a few clicks. In my experiments, I observed that the AI effectively regenerated detailed areas, typically in a way that would go unnoticed by the average viewer, especially when displayed in smaller formats on a screen. This capability is truly impressive!

 

Extending an Image

The capability to extend an image using Generative Fill is truly remarkable. This AI technology generates new portions of an image based on the existing scene and the provided prompt.

In my initial tests, the AI performed admirably when extending scenes with straightforward subject matter, such as a blue-green reef or a field of hard coral. I experimented with image extension to adjust the original photo’s aspect ratios for social media purposes, such as converting a 3:2 ratio to 4:5, as well as creating elongated formats for underwater photographs.

For years, adept graphic designers have manually extended underwater images, though their ability to clone larger areas while maintaining a natural appearance was limited. The introduction of Generative Fill empowers a broader audience to achieve realistic results that are suitable for social media and small web applications with ease.

 

AI and the Future of Underwater Photography

The topic of photo editing has long been a point of contention among photographers. Is it considered cheating, or is it an integral part of the artistic process? Should a photograph solely reflect what was captured by the camera, or is it permissible to enhance or modify the image during post-processing to realize the photographer’s creative vision? Notably, even the esteemed photographer Ansel Adams employed dodge and burn techniques in his darkroom.

With the advent of Adobe’s AI tools, significant photo modifications have become accessible to editors of varying skill levels. We are already witnessing a blend of reality and artificial intelligence in social media content. Furthermore, there have been instances where photographers have utilized AI to secure victories in major photography competitions before eventually disclosing their methods.

While some AI-generated images can appear remarkably realistic, this particular depiction of a sea lion diver, although entertaining, deviates significantly from authenticity.

 

AI image created in Midjourney

Everyone is entitled to their perspective. Personally, I believe that photography will continue to evolve, splitting into two distinct groups: those who relish the act of capturing images and those who create art through a fusion of photography and AI-enhanced post-processing.

Professional photographers across various fields, whether in fashion, sports, or natural history, are approaching a critical juncture. Fashion photography is likely to be the first sector to be significantly influenced by AI technologies. While sports photography presents unique challenges, advancements in AI and exceptional autofocus capabilities may eventually enable automated cameras to replace human sports photographers. Conversely, natural history photography may endure longer, as AI currently lacks the extensive reference material needed to recreate the rare scenes that we, as humans, encounter and photograph.

My concern is that as sensationalized images proliferate, we risk rendering real-life experiences, such as scuba diving, mundane for younger generations who expect AI-generated visuals. It is imperative for us, as divers, to embrace the remarkable technology available to us while also encouraging a return to authentic experiences—putting away our devices, taking deep breaths, donning our masks, and immersing ourselves in the wonders of the deep blue ocean.

 

AI Input Datasets Source

Coral Mountain Data is the pioneering underwater data mining company located on Phu Quoc Island, Vietnam. We specialize in generating underwater datasets designed to enhance artificial intelligence models. Our offerings include high-resolution underwater photographs and videos, capturing a diverse array of scenarios to enrich the training and performance of AI models.

 

Coral Mountain Data Factory is located in South Phu Quoc Island with professionally trained & skilled employees to produce the best datasets.