In the past few years, we have seen a real influx of legitimately useful AI-powered image-editing tools and programs, and their presence has brought about a bit of controversy regarding their usage and future place in the industry. Will the future of post-processing be powered by AI?
Coming to you from Blake Rudis with f64 Academy, this thought-provoking video discusses the future of AI-powered photography editing. Programs like Luminar 4 and PortraitPro have brought features like AI-powered sky replacement and skin retouching in the past few years, and having used both of them, I have to admit that I have been really impressed, and both applications have been integrated into my workflow. Regarding the future of post-processing, I think there will always be room for a personal creative touch, as that is why the majority of us are in photography. Where I think AI will see a larger role is in the sort of repetitive tedium that really isn't fulfilling but that still needs to be done: things like retouching large batches of corporate portraits or basic global edits on a huge collection of wedding photos. For example, just to test, I ran a batch of about 50 portraits through PortraitPro, and the results were highly impressive.
What do you think the future is? Check out the video above for more.
Back in the early 80's even before I pickled up a camera of any kind, analog or digital, my high school physics teacher once said, "If it works, it's not AI." I have not seen anything yet to argue the point.
Could not agree more. Blake’s assessment is spot on.
The future is all professionals being replaced with interns pushing AI buttons. I bet you're laughing at that comment right now, aren't you?
Just an observation. Not just of editing but of image capture probably. Onboard the camera itself.
I don't agree. AI, or more correctly Automation in photography is here to stay. Autofocus on a camera anyone? How many photographer prefer manual focus to autofocus? AI is just another tool in photography and we are just at the early stages of it been used in photo processing. Look at the processing a mobile phone does without any intervention by the user. It is not great but it is much better than a "straight" unprocessed image from the camera.
For another example take Luminar's sky replacement or Photoshops selection tools, which do a fantastic job of selection. You might get a better selection with the pen tool but if you are editing 1000's of images for weddings which tool would you chose?
AI has its uses but not yet for the interpretation of your vision but sometime soon it just might...
Software is a tool, not an obstacle because you don't like an "auto" based feature. Maybe AI in the future will write better blog articles, definitely poorly done.
Despite my first post dated Sept 19 2020, I do agree with all of your post. But my point is not about the automation of photography or post processing, but rather what AI is exactly. All the automation settings programmed or developed for in camera settings does not make it AI. True AI is machine learning and comprehension without the assistance of man. No programmed algorithms, no built and developed sensors. Straight forward self awareness.
Who is AI been marketed too? Is it for the fine art print producer? The one that wants to tease out exceptional images from their photos? Or the millions of average photographers that want to improve 30 pictures they took of the kids sports event, family picnic etc. Its the later of course. If the AI, Machine learning or just algorithms can improve images straight from the camera pretty quickly that will meet many peoples needs. Some it will fail on but many will be improved (if not by this generations algorithms then by future generations). For those that want to make exceptional images they will still have to put the work in and learn the process.
The downside of AI, Instagram filters and other automation processes is that the end result can be a bit "samey" and they all have the same "look".
As others have mentioned it is not clear when a combination of machine learning, mathematical algorithms, deep learning and computer vision combine to form a true (specialized) AI solution. If it can take most images from across the range of subject matters, lighting, color space etc and consistently produce pleasing images then its getting there.I suspect that these new algorithms will work reasonably well with range of common images. Pictures taken in daylight of people and simple landscapes etc.