Sora is dead and camera manufacturers agree – the photo news of week

Contents

Sometimes the tech industry cleans up faster than it can create chaos. OpenAI, the company that just a few months ago wanted to revolutionize the creative world with its AI video generator Sora, has unceremoniously scrapped the tool this week. Not quietly, not gradually – but with a social media post that reads like a farewell speech at a company party where no one is really sad.

From Wunderkind to Side Quest

Sora's story is as short as it is instructive. Sora 2 was launched with great fanfare in September, and it's over in March. Not only is the app disappearing, as the Wall Street Journal reports – the entire project is being buried. CEO Sam Altman himself informed his employees that OpenAI will no longer support products based on generative video models. Even the video function in ChatGPT is on the chopping block.

The reason? OpenAI is apparently preparing for an IPO and wants to focus on what actually makes money: productivity and enterprise applications. Fidji Simo, head of applications, had already warned last week that they needed to stop being distracted by “side quests.” Sora was apparently exactly that – a side mission that consumed more resources than the market demanded.

Particularly spicy: Disney had agreed as part of an arrangement to invest one billion dollars in OpenAI, including licensing popular Disney characters for generative AI apps. With the end of Sora, this deal is also off the table. Disney was diplomatic: they “respect the decision,” but a billion dollars is also respected. We reported on Sora's creative potential, now one may ask: Was the artist Boris Eldagsen with his experimental Sora video in the end one of the few who could actually get something meaningful out of the tool?

Generative AI Doesn't Belong in a Camera

While OpenAI is thus burying its AI video ambitions, the major camera manufacturers have taken a remarkably unified stance at CP+. Canon, Nikon, Sony, Fujifilm, OM System, Panasonic, and Sigma agree on one point where little unity would be expected: generative AI has no place in a camera.

This is, one must say, possibly the first time in the history of the photo industry that all seven major brands hold the same opinion. Sigma CEO Kazuto Yamaki puts it succinctly: “Photography is not just about the result, but about the joy of the process.” If it were just about the image, AI-generated images could replace photos – but that is precisely not the case.

Nikon emphasizes the human factor of creativity and points to the risks that generative AI poses to the concept of originality. Canon warns of fakes and calls for better authenticity proofs. Fujifilm wants to be able to clearly distinguish whether an image was generated or photographed. OM System sees AI as a tool – for noise reduction or image composition, for examplebu, t not as an image generator. Panasonic emphasizes that the subject must be real, especially in video productions.

Mind you: no one is fundamentally rejecting AI. Autofocus detection, subject tracking, noise reduction – all these processes benefit from machine learning, and manufacturers use it extensively. The red line is drawn where something is created out of nothing, where it is no longer photographed but fabricated. The fact that smartphone manufacturers, of all people, are increasingly staging their product launches around precisely this capability is met with unanimous head-shaking by the camera bosses.

Canon Relies on AI – But the Right Kind

That the rejection of generative AI by no means means technophobia is shown by Canon in a separate conversation at CP+. Go Tokura, Executive Vice President and Head of the Imaging Group, even sees deep learning and AI as the engines for the next major breakthroughs in camera technology.

The discussion is about AI-assisted white balance that uses subject recognition, AI-based noise reduction in software, and digital lens corrections that improve image quality with optical measures. Particularly exciting: Canon sees potential in resolving the eternal tension between high resolution and high speed through AI-based image processing. The EOS R1 already offers in-body AI upscaling, and Canon also hints that future products could combine both – many megapixels and high continuous shooting rates.

“Software can help to a certain extent,” says Tokura. “Through deep learning technology, we can improve this area.” This sounds less like a revolution and more like consistent evolution – but that is precisely what photographers expect from their tools: to get better without betraying the basic idea.

Does Photography Still Need Professionals?

Less diplomatic, but all the more provocative is an essay on Fstoppers. Finnish photographer and author Alvin Greis asks there whether photography still needs to be treated as a profession – and his answer is, to put it mildly, not what professional photographers would like to hear.

His thesis: Professional photography historically existed primarily as a form of risk control. Expensive equipment, limited knowledge, no immediate feedback – all this made a professional necessary, not because the tasks were so complex, but because there was no other way to achieve a usable result. The barrier to entry was higher than the actual requirement. Professional photography, according to Greis, “occupied far more space than the tasks actually required.”

The comparison with the typist profession is as apt as it is painful: typing was once a profession – not because writing required specialists, but because errors were expensive and difficult to undo. As soon as keyboards, undo functions, and word processing eliminated the risk, the profession disappeared. Greis sees photography on a similar path.

Of course, the thesis has its limits, and the comment section on Fstoppers clearly shows this. Wedding photographers point out that there are no second chances at a wedding. Portrait photographers emphasize that trust and people skills cannot be replaced by an app. A retired portraitist from Toronto reports that he is inundated with requests because the results from new photographers and AI are simply “over-processed.”

Greis admits that professional photography continues to exist where it is actually needed – at weddings, in complex productions, wherever repeatability and reliability are required. But he insists: this is a narrow area, not the broad mass of photographic tasks.

One does not have to agree with the essay to find it valuable as food for thought. Because the question of what makes a photographer a professional when the technical barriers fall is a valid one – even if the answer is more complicated than an economic model suggests. If you're looking for controversial reading over the weekend: the full essay on Fstoppers is worth it, if only for the comments.

The Week in One Sentence

OpenAI learns that not every AI needs a product, camera manufacturers learn that unity is possible, and the photo industry learns – once again – that the most exciting questions are not about the next camera, but about the meaning of it all. So: Go out and take pictures. With a real camera. The pixels are real, the emotions are too – and that's more than any AI can claim for itself.

(tho)