Meta ships preset-prompt video editing before text prompts, across three surfaces

The interface is a menu, not a text box

The most telling detail in this launch is what users cannot do yet. Meta says you upload a video and then "explore more than 50 editing prompts" to transform 10 seconds of footage. The examples are fixed scenarios: turn your clip into a graphic novel, restyle it as a vintage comic illustration, or convert it into a video game with fluorescent lighting and battle clothing.

Open-ended text prompting is explicitly held back. Meta writes that "later this year, you'll be able to edit videos alongside Meta AI with your own text prompts." So the first release is a curated palette. That is a deliberate frontend decision, not a limitation of the underlying research — the Movie Gen work it cites already accepts "simple text inputs."

Once you've chosen a preset prompt, Meta AI will edit your video to match the selected scenario.Montana Labs

Presets do real product work here. Each one is a prompt the team has already validated against the model, so the output stays inside a known-good range. A free text box exposes the full failure surface of a generative video model; a menu of 50 tested transformations does not.

Constraints that keep costs and expectations bounded

Two numbers frame the release: over 50 prompts and a 10-second edit window that is "free for a limited time." The short duration caps per-edit compute and sets a clear boundary on what the feature promises. Video generation is expensive, and a 10-second ceiling is as much an economic control as a creative one.

The "free for a limited time" language also signals that pricing is coming. By launching with a fixed clip length and a fixed set of transformations, Meta gives itself a clean unit to eventually meter. That is easier to reason about, and to bill for, than open-ended edits of arbitrary length.

Three entry points, distribution baked in

The feature ships simultaneously in the Meta AI app, the Meta.AI website, and the Edits app. Where you can publish depends on where you edit: from Edits and the Meta AI app you can post directly to Facebook and Instagram; from the Meta AI app and Meta.AI web you can post to the Discover feed.

Placing the tool inside Edits matters. Meta says the feature "is built into the Edits app" so it "can become a seamless part of the creative process," and that it worked with creators to choose which prompts would resonate with their audiences. That is the frontend strategy in plain terms: the generative step is embedded in an existing editing workflow, and the output has a one-tap path back into Meta's own feeds. The edit and the share are the same motion.

What the preset-first rollout implies

For teams building consumer-facing generative features, this launch is a concrete example of sequencing. Meta had the text-conditioned capability from Movie Gen but chose to expose it first as a bounded set of presets, saving open text prompting for later in the year.

The presets serve three purposes at once — they constrain output quality, they cap compute per edit, and they give non-expert users an obvious starting point instead of a blank field. When the free-text version arrives, Meta will be adding a harder-to-control interface on top of a product whose reliability expectations have already been set by the curated menu. Shipping the safe surface first, then widening it, is the pattern worth noting.