The rise of AI in Graphics
Throughout the 1980's and 90's there was a narrative that 'computers and robots will replace our jobs'. Thirty years later it is apparent that this never happened, instead we’ve just changed the way we work. Office workers no longer trawl through hundreds of filing cabinets and thousands of paper files to get the information they need, today the files automagically appear on screens in front of them. A somewhat analogous change is now happening in the graphics industries. Graphic design, video post-production, Game development, photo editing, film CGIare some of the contemporary jobs that are undergoing a quiet revolution, the basic job is the same, but how it's done is becoming radically different.
The use in 2D Graphics
Over the past two years the use of AI in graphic design and photo editing has gone from the experimental stage to mainstream. Adobe machine learning modules have found their way into most of the Adobe CC products of 2019/2020, where they can auto-arrange items, auto-crop, fill blank areas with content made up (dreamed up) by the AI, the AI face editing function can change the shape of eyes, mouth, make the subject smile or frown, all by moving a few sliders. It's clear that these features are the start of a trend, and that Adobe will be adding new AI derived functions with every update.
While Adobe is the undisputed market leader in the world of 2D graphics software, new software houses are posing a serious challenge to their supremacy. In 2018 Topaz Labs released Gigapixel, a piece of software designed solely for upscaling images, by using AI to work out where object boundaries are, it can take a small blurry image and output a larger sharp image. Had you suggested this even 5 years ago most people in the tech industries would have laughed and said it's not possible. The extra pixels added to these images are not 'real', in the sense that they do not represent what was really there a the time of the photograph, the AI is guessing what the details and texture were, and adding something that looks real. Topaz Labs several other products that use AI to denoise, mask, sharpen and otherwise alter and improve images.
AI use in Video
Deepfakes were big news in 2019, these videos of celebrities and politicians saying and doing unexpected things made the world a more complicated place. How do we know that a politician really said those words? At the time of writing creating a deepfake video still requires significant skill and effort, but the process is continually being made easier, and we probably won't have to wait for too many versions of Adobe Premier before it becomes a standard feature.
While Adobe premier doesn't currently (at the time of writing) have a deepfake function, it does use AI in several of its new features, such as Auto-framing, where the subject is kept centred in a cropped or resized video. Premier has introduced AI colour grading that identifies skin tones and keeps them realistic, while altering the overall colour cast of the video. Premier will now also create fake slow-motion video where the intermediate action frames are made up by AI rather than just by blurring frames together as happens currently.
AI use in CGI
The film industry has always been at the forefront of new technologies and AI is no exception, the uses today include scanning actors faces and recreating them on digital characters, so bringing dead actors back to life, or making them look younger or older.
Many of the tedious tasks associated with making CGI effects are being taken over by intelligent algorithms, for example, the insertion of a CGI object in a real world scene would require the artist to match the lighting of the CGI object for every frame, but now this can be automated with AI. The pasting of faces onto CGI bodies would have traditionally meant the VFX artist would have to recreate the facial movements of speech, and expressions by hand, slow and difficult process, and the result would be ultimately unconvincing, but today AI software will do this with little or no effort, and in a much more realistic way.
The takeaway from all this must be that the role of the graphic artist is not going away any time soon, but the way they work is likely to change significantly over the next decade.