By now, you’ve heard the statements, “Photography is dead,” “Hollywood is dead,” and “Creativity is dead.”
This is really just a ridiculous idea because, with AI, the Arts will only get better.
Photography and the Arts, in general, will always be about the moment and the human experience. Whether that’s landscape photography, photojournalism, or parents recording their children taking their first steps, AI can’t and won’t ever be able to replace these things, and Hollywood will always be about storytelling.
Photography is an art and a science. It is also a field constantly evolving with new technologies and innovations, and AI is a new and powerful tool within this space.
*In this article, I’m sharing the AI artwork by my friend Chris Do, who uses Midjourney V6. I use Stable Diffusion and can’t quite get the detailed images this new MJ6 offers so that I will feature his content for now.
[toc]
3 Ways AI Will Change Photography
AI has the potential to enhance photography in many ways, such as:
1. Improving the quality and the creativity of the images.
AI can help photographers edit and enhance their photos with advanced tools and algorithms, such as noise reduction, color correction, exposure adjustment, and artistic filters.
AI can also help photographers generate new and original images or ideas from scratch, greatly boosting pre-production efficiency.
Even with where we are today, there is no longer any excuse for photographers and cinematographers to ever go into a shoot with a half-baked concept. Preproduction and planning new ideas is so fast and easy to do now.
As a fashion photographer, studio cinematographer, or art director, you should be eager to master this technology because the future belongs to those who can use it well. But you must also balance your AI skills with your artistic vision, which requires constant learning and improvement. This means there is more you have to learn if you want to stay competitive.
2. Replacing Some Of Our Cameras And Accessories
AI can enable photographers to take photos with devices other than traditional cameras, such as smartphones, tablets, and wearable gadgets. It can also simulate the effects of different lenses, apertures, and lighting conditions without physical equipment. AI can also help photographers capture images beyond human vision’s capabilities, such as infrared, ultraviolet, and X-ray images.
We’re not there yet, but maybe we’re only about 10 years away from what I think will happen.
What AI Will Do To Our Cameras
Your need to capture the most noise-free, high-resolution, perfectly sharp image will disappear. You’ll feed the software an image that represents what you want, and it will go through a denoise and uprez process to enhance all the details on trained data.
This means it could perfectly add detailed eyelashes or an extremely shallow depth of field. Color out all the noise and paint in perfect detail. Midjourney is already doing this with nothing but text prompts, and you can already feed images into stable diffusion, and it will re-render them in a similar likeness.
The problem today is the computational and data transfer bottleneck. Uprezing a 1024×1024 image to 2048×2048 with my Geforce 4070 takes about 4-5 minutes with all the goodies turned on. Trying that on a 6k image would probably take 30 minutes to an hour on today’s (2024) Nvidia hardware. We are quite a ways away from being able to do this in real time. And even offloading onto some Apple or Samsung server with a smartphone would be too much of a resource bottleneck for these companies with how many people are using their phones today. Although I could see a subscription AI enhancement service in the future, where you turn your phone in AI mode and take a shot, it sends it off to the cloud to have it enhanced with your set specifications.
That would be cool.
This could simulate f1.2 lenses, shallow depth of field, and have the best sharpness you could ever desire. Of course, it wouldn’t be a perfect representation of reality, but it doesn’t need to be.
Feed it an image of a forest and return something like the image below. We do this with photo editing anyway, so why not make it instant? As far as I know, some companies are trying to do this already, but they don’t have a diffusion neural network model that powers stable diffusion and mid-journey.
Large Sensors, Expensive Lenses Might Go Away
I’m talking in 10-15 years, but a 1:1 integration with real-time AI uprezers will eliminate the need for fancy gear.
And I’m not talking about Topaz Gigapixel. While Topaz is great, it sharpens algorithms and turns things a little illustrative. While good, Topaz and its Gigapixel tech are all designed to do their job quickly. It’s consumer-facing, so it has to be functional, easy, and quick.
3. Helping us learn and discover new things.
AI can help photographers analyze and understand their images with features such as object detection, face recognition, scene classification, and semantic segmentation. We’re getting there, but cameras can still do a lot more.
AI can also help photographers find and explore new subjects, locations, and perspectives using features such as image search, recommendation systems, and augmented reality.
What You Need To Do As A Photographer
This is the best part—if you just like taking pictures, you don’t need to do anything. Most companies like Apple or Samsung will drip-feed us features over the years. Big camera brands like Canon, Sony, Nikon, and Fujifilm will continue to enhance their cameras’ AI features, making them smarter, more accurate, and more capable.
I always get stuck believing, “I have a Nikon Z8; I’ll never need another camera again.” In terms of image quality and what I like today, that’s probably true. But so much is changing, and it’s changing fast.
It’s crazy, but I think of all these cameras; Fujifilm is the only one on the right track with how it allows us to use film simulators and processes the data intelligently to make those images. This has been huge for Fujifilm, and hopefully, it will inspire other brands to follow.
