For years, OpenAI’s Sora was hailed as the breakthrough poised to redefine video creation and social media. It launched to massive fanfare in December 2024—but just last month came the announcement: it’s shutting down at the end of April. Chalk one up for the human race.
There are several reasons for the closure. In short, it didn’t generate enough revenue to justify the overhead, copyright issues, and dealing with all the strange content that was popping up. It made sense to refocus resources on things like physical AI.
We experimented with Sora 2 Pro extensively for the past year. The app targeted online social content creators, but we were intrigued by the possibility of its cinematic qualities. Could we create amazing Hollywood-quality commercials for under a thousand dollars? We experimented for months, even producing a few commercials.
At the end of the day, when we heard the news that it was closing, we weren’t very surprised and definitely not disappointed.
Here are some takeaways from our experience.
Generating AI videos can be frustrating
GenAI is good enough for simple generic b-roll type scenes. For example, it’s fine for things like a hike in a forest or a person listening to music on a phone. But when you get more specific, then it becomes trial-and-error. In one scene, a doctor is in surgery and gives a quizzical look at what he’s seeing. I assume because Sora is geared toward social content creators who want finished videos fast, it often added unnecessary lines of dialogue, the wrong look, and/or cuts to different shots. We tried rewriting and simplifying the prompts. We even tried creating reference images. Nothing worked well.
We could have shot this doctor scene in about an hour with a good actor. Of course, it would cost thousands of dollars to do it, but at the end of the day, we would have served the script.
Here are stills from those failed generations. Quality looks cinematic but we weren’t getting the facial expressions and actions we wanted.
Sora 2 Pro Pricing Structure wasn’t optimal
To access Sora 2 Pro, we had to purchase the ChatGPT pro subscription at $200 per month, which obviously didn’t break the bank. But it only provided 100 credits per day, and for most of our subscription, it required 10 credits per high-quality video. Needless to say, we burned through those credits very fast and the budgets didn’t justify purchasing more. Because of this limitation, we spent several days on that one doctor scene until we finally gave up and rewrote the script, and that wasn’t too successful either.
The credits also reset on a 24-hour clock, which meant that if we spent the credits at 5 PM, then it wouldn’t reset until 5 PM the next day. Our schedules vary every day so we couldn’t always work on these projects at the same time during the day, which ultimately led to lost credits. In contrast, Adobe Firefly’s credits reset on a calendar month. I’m not sure why OpenAI thought they could monetize better on a 24-hour clock rather than just renewing credits at midnight.
What about the other GenAI tools?
We actually started with Adobe Firefly, switched to Google VEO before investing full time to Sora. We already had a full subscription to Adobe Creative Cloud so Firefly made sense, but it was terrible. Recently, they added third-party apps, which made it a lot more practical. I assume since Firefly is packaged with CC and not a stand-alone subscription, it won’t go away soon and may improve over time.
Google VEO works well for photo-realistic videos, but it has the same problem as the other GenAI tools. It takes repeated generations to get something close to what you want. The main reason we switched to Sora was that VEO wasn’t as cinematic.
Sora also provided better tools like the Extension function, which made it easier to expand a scene.
Overall, neither Adobe nor Google will be pulling the plug on GenAI video creation tools.
GenAI won’t die anytime soon.
Sora’s closing does not mean there was a GenAI bubble. It revealed that the hype was greater than reality, and perhaps there was a rush to market. The technology still has some ways to go. It also showed that monetizing this technology may require an innovative approach. But we believe it will get there.
Picturelab is a full-service video production company – we shoot videos the old way and we also use Generative AI tools. Contact us to learn more about everything we do.