New AI Video Editor - Text to Video is Mindblowing!

Runway’s upcoming “text to video” pitch is landing less like a brand-new video generator and more like a fast, prompt-driven AI video editor—where text controls edits such as color grading, object removal, inpainting, and green-screen-style masking. The viral demo that circulated online shows a “magical box” workflow, but the early segments appear to import existing footage and then apply AI-assisted adjustments rather than synthesizing entirely new scenes from scratch. That distinction matters because it sets expectations: the near-term breakthrough is editing automation, not fully generative cinema.

In the demo, prompts like “make it look cinematic” trigger changes that resemble automatic color and look adjustments on an existing clip. Other prompts point to more advanced manipulation. “Removed this object” suggests the system can identify a target and eliminate it from the surrounding frames—an operation that typically requires heavy processing to maintain visual consistency over time. The transcript also flags a key nuance: some prompts use “import” (implying footage is brought in and then edited), while later prompts use “generate” (implying AI generation at least for still imagery). When “generate a lush Garden” appears, the results shown are still images presented across multiple options, not moving video.

The most eye-catching editing capability shown is “green screen character.” Instead of requiring a traditional chroma-key setup, the system performs masking and background replacement-like compositing even when the subject isn’t originally shot against a perfect green screen. The demo includes controls such as feathering and masking, and it previews a replacement clip over the masked area. The workflow then extends into a browser-based editing experience on Runway’s site, where AI-powered tools are already available.

On the Runway website, the editor is presented as “edit video in seconds” with features including background removal and inpainting. A live test described in the transcript deletes a moose from a scene using an inpainting brush. The result is described as surprisingly effective, with the system tracking the subject’s motion and handling difficult edges—though not perfectly, as shadows and artifacts can still shift during playback. A second test uses a dancer clip to demonstrate green-screen-style extraction and replacement, again described as performing well despite the subject not being shot on a true green-screen background.

Overall, the takeaway is that Runway’s “text to video” direction is already materializing as AI-assisted video editing: prompt-driven effects, object removal, inpainting, and compositing. If full text-to-video generation arrives later, it will likely build on these editing primitives rather than replace them outright. For creators, the practical impact is immediate—fewer manual steps for masking, cleanup, and background swaps—while the longer-term promise is turning natural-language instructions into end-to-end video changes.