Alright, let’s get this heresy started. I’ve been running the Midjourney + manual editing (After Effects, Runway, you name it) pipeline for client work for over a year. When Sora access finally landed, I did the logical thing: I migrated a whole slate of upcoming explainer and social clips over to it, convinced I was about to shave 50% off my production time. Two weeks and a frankly alarming number of credits later, I’m crawling back to my old stack. And not because Sora is "bad"—it’s astonishingly good—but because it’s a classic case of raw capability versus practical utility.
The hype is all about the *moments*: the perfectly simulated light, the physics of a fabric, the breathtaking establishing shot. And yes, generating that single perfect shot is faster than ever. But try to build a consistent 15-second sequence with a specific character, maintaining wardrobe, setting, and camera angle. You can’t. You’re playing a lottery every time you prompt, and the house always wins. My benchmark was a simple scene: a person in a distinctive striped sweater walks from a kitchen to a living room and picks up a book. Midjourney gave me a stable character I could then animate consistently using other tools. Sora gave me eight different people, three different sweaters (when it remembered stripes at all), and a living room that changed architectural eras between cuts. The "editing" I was saving? It just moved from After Effects to the prompt box, becoming a more abstract and expensive form of finessing.
Then there’s the control problem, which is really a predictability problem. With my old stack, I knew where the costs were: a Midjourney fast hour, a Runway Gen-2 render, my own hourly comp work. It was linear. With Sora, the cost is in the endless iterations to *approach* what you want, with zero guarantee of convergence. It’s like negotiating a SaaS contract where the service level agreement just says "we’ll try our best." For commercial work, that’s a non-starter. You can’t budget for "maybe 20 generations, maybe 200."
And let’s talk about the vendor lock-in. With a modular pipeline, if one tool stagnates, I swap it out. If a better image model drops, I plug it in. If a new animation tech emerges, I test it. Sora is a magnificent walled garden. You get what they give you, at the pace they decide, for the price they set. There’s no sub-vendor comparison, no open-source licensing alternative on the horizon for this tier of video. That should make any procurement-minded person nervous.
So I’m back to stitching together my Frankenstein’s monster of tools. It’s messy, it requires actual skill beyond prompt-craft, and it’s decidedly un-sexy. But I can deliver it on budget, on time, and with a consistent output that doesn’t leave me explaining to a client why the protagonist mysteriously changed ethnicity halfway through the video. Sora is a glimpse of an incredible future, but for reliable, billable work today? It’s a staggeringly beautiful toy.
—Bella
Price ≠ value.