
YouTube launched Reimagine on March 18, 2026, and it does something none of the other short-form video platforms have tried at this scale. It takes a single frame from an existing YouTube Short and generates an entirely new 8-second video from it, using Google's Veo video generation model and Gemini for text prompts. The feature sits inside YouTube's existing Remix tool, although what it actually does is closer to AI-assisted content creation than anything the word "remix" usually implies.
Key facts at a glance
A viewer watching a Short taps Remix, then selects Reimagine, and YouTube lets them pick a single frame from the original clip. From there, they can upload up to two photos from their camera roll, usually of themselves or a prop, while Gemini suggests text prompts to guide the output. The user can also write their own prompt, and Veo then generates a new 8-second video with audio based on the frame, the uploaded references, and the prompt.
What separates Reimagine from existing AI video tools is how tightly it is woven into the consumption flow. You do not need to leave YouTube, open a separate app, export a frame, or paste it into an AI generator, because the entire creation step happens inside the same screen where you were watching content. That compression of the gap between watching and creating is the real product decision here, not the AI model itself.
YouTube built a direct link between every Reimagined Short and the original video it was derived from, so tapping on a Reimagined Short takes you straight back to the source clip. The credit is visible and structural, not a footnote buried at the bottom.
AI remixing raises an obvious question about ownership, though. If someone takes your frame and generates a new video, is that a derivative work or a new creation? YouTube is sidestepping the legal debate by treating Reimagined Shorts as a distinct format that always carries attribution, similar to how duets and stitches work on TikTok but with the added complexity of AI-generated imagery.
Whether creators are comfortable with this arrangement is still an open question. YouTube tested Reimagine with a small group from late February before launching it broadly, and the early response has been mixed. Some creators see it as free distribution, while others are uneasy about having their frames turned into AI content they did not approve.
TikTok, Instagram, and YouTube have all been shipping AI creation features in 2026, but YouTube is the first to embed generative AI directly into the remix flow. TikTok's AI tools sit mostly on the commerce and editing side, and Instagram's Restyle works more like a filter, whereas Reimagine produces entirely new video from a single input frame.
If Reimagine drives measurable engagement on Shorts, expect TikTok and Instagram to build their own versions. The underlying models already exist, so the question has always been where in the product experience to put them, and YouTube answered that by placing generative AI at the point of remix rather than at the point of original creation.
Before Reimagine, creating a Short required having footage, whether that meant shooting something, screen-recording something, or editing clips together. Reimagine lowers that bar to a single frame and a text prompt, which is a meaningful reduction in effort for viewers who never posted before. For brands and content teams it opens a new format category, although it also raises a question about how much AI-generated Shorts content YouTube's recommendation system will actually distribute compared to original footage.
YouTube has not said publicly how Reimagine content will be treated by the algorithm, and whether AI-generated Shorts get the same recommendation weight as original uploads will likely determine whether Reimagine becomes a mainstream format or a novelty.
