• Sora can now turn pets and objects into characters
• We took the new feature for a test drive
• Matching the right prompt with your desired output is still challenging

OpenAI has given aspiring AI filmmakers a potentially powerful new feature within the Sora app. The company has expanded the Cameo tool for making you (or at least your digital avatar) the star of your AI videos to pets, toys, and even dinnerware. The new Character Cameo feature lets you film animals and objects around you and make them into AI constructs the same way you do for yourself, then place them into Sora’s AI-generated videos as reusable, animated “characters.”

Videos made with Sora 2 are a mixed bag, no matter how clever you get with prompts. So while the popular Sora app is full of people showing off their Cameos and Character Cameos, I decided to join in.

Creating a Character Cameo is easier than making one for yourself. You just need a few seconds of video of the subject, ideally with good lighting and nothing in the background. You then upload it to the Character Cameo tool, where Sora comes up with a name, a tag, and even a whole elaborate capsule summary of the personality it thinks the object should have, but you can edit it to your heart’s content before publishing.

Once published, the character is available for tagging in Sora prompts, and you’ll see it in the resulting video, barring accidentally tripping a rejection to a request over copyright or because your “pet cat” is actually a human.

Costume Pup

The first test was with my dog Cabbage, using a video of her running around. She’s quite agile, so I decided to have some fun and make her a show-off dancer. For no reason other than to see what would happen, I had Sora set her performance in the Antarctic in front of a bunch of penguins. I only specified she wear a “fun dance outfit.” The AI decided that meant the Ice Capades and bangles.

It’s a cute video, but before I moved on to the next Character Cameo, I couldn’t resist making Cabbage an ersatz superhero, requesting a cosmic look like I did when testing Meta’s new restyling tool. I tagged Cabbage’s avatar and asked that it glow and accelerate into space. Though hardly without flaws, I was impressed with both the costume changes and the stage direction. The scenery was much less realistic, considering I specified New York City. That may not have been enough detail for Sora, though.

Living music

For the second Character Cameo, I first picked an Elmo-branded backpack my son owns, but that was too copyright-specific, so instead I chose this toy piano/DJ set. I asked for a video of it performing a song on its own. At first, that led to nightmare fuel-style versions of the toy. If it looked like the one below, I would be calling an exorcist before going near it.

Still, I persevered, and with more specific instructions about what was playing and singing, I ended up with the video above. It’s flawed, with the toy itself not looking quite right, and the dog DJ on the turntable turned into that odd child, but for two minutes of processing, I didn’t have many complaints.

Goblets up

For my third character, I went with an elaborate kiddush cup from my wedding. Its pomegranate-vine theme and hammer-beaten metal look made me think of Indiana Jones, so I went with a prompt that fit. I tagged the goblet and asked for a video of it in a mysterious temple where wine from nowhere would fill the cup and an unseen voice would ask if the viewer was thirsty.

This video seemed the most flawed to me, no matter how many variations I came up with. The physics of the magical aspect seemed to confound Sora. After a handful of failures, the video above at least mostly got it right.

Long-term cameos

As with the Cameo you can make of yourself, there’s a lot of potential long-term value for Character Cameos. Once uploaded and tagged, you can reuse it across multiple videos in any way you see fit. You can keep it to yourself, or try to make your pet (or cup) famous by sharing it around.

But, like Sora as a whole, you’re not likely to be satisfied with the first prompt result, which can be frustrating. And while I applaud the ban on copyrighted characters or realistic likenesses of people, it is a bit overly strict, to the point where even a poor doodle drew a warning about trying to make a person a Character Cameo.

For now, the feature helps make Sora more fun and interesting, but OpenAI will need more if it wants people to treat it as fount of new movie stars, no matter what my dog thinks.

Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds. Make sure to click the Follow button!

And of course you can also follow TechRadar on TikTok for news, reviews, unboxings in video form, and get regular updates from us on WhatsApp too.

Eric Hal Schwartz is a freelance writer for TechRadar with more than 15 years of experience covering the intersection of the world and technology. For the last five years, he served as head writer for Voicebot.ai and was on the leading edge of reporting on generative AI and large language models. He’s since become an expert on the products of generative AI models, such as OpenAI’s ChatGPT, Anthropic’s Claude, Google Gemini, and every other synthetic media tool. His experience runs the gamut of media, including print, digital, broadcast, and live events. Now, he’s continuing to tell the stories people want and need to hear about the rapidly evolving AI space and its impact on their lives. Eric is based in New York City.