Turn a text prompt — or a still image — into a short AI video with synced audio, right in your browser. Free, no sign-up, no install. The engine starts on demand and runs on GPU infrastructure. A free alternative to paid AI video tools like Kling, Sora and PixVerse.
Describe subject, action, camera and mood. "A red sports car drifting on a wet street, neon reflections, low angle" beats "a car".
The engine handles short, single-shot scenes best. For multi-shot stories, generate several clips and stitch them.
The first launch may take up to a minute while the GPU warms up. After that, clips render far quicker.
No timeline editor to learn, no assets to upload. The prompt is the interface.
Write the subject, the action, the camera angle, the lighting and the mood. More detail gives the model more to work with.
Press generate and the engine renders your clip on GPU hardware. A typical generation finishes in under a minute once warm.
Preview the result in the panel, then download it. Re-roll with a tweaked prompt if the first take isn't quite right.
Pure prompt-driven generation. Describe the scene; the engine renders a short clip.
Add a first frame to bring a still image to life with motion — wind, parallax, camera moves.
Vertical clips for Reels, Shorts and TikTok — set the aspect ratio in the engine.
The video engine works two ways, and knowing which to reach for saves a lot of re-rolls. Text-to-video starts from a prompt alone — best when you want the model to invent the whole scene, from the subject to the camera move. Image-to-video starts from a still frame you provide, which gives you far more control over how the shot looks, because the composition, colours and character are already fixed; the model only has to add motion.
For either mode, describe one clear scene rather than a sequence. "A paper boat drifting down a rain-filled gutter, slow motion, shallow depth of field" works; a prompt that tries to cut between three locations will not. Include the action (what moves), a camera cue ("slow push in", "wide shot", "drone shot") and the mood or lighting for the strongest results.
Keep expectations realistic: this is a free, open model, so clips are short and single-shot. For longer content, generate several clips and stitch them together in any editor. The engine runs on shared GPU infrastructure and starts on demand, so the first render after a quiet spell can take up to a minute — after that it's quicker.
Need a starting frame? Generate one with the free AI Image Generator first, then bring it here to animate.
Yes — free with no account, supported by ads on the page. Because it runs on shared GPU infrastructure, it can queue at busy times.
The engine produces short, single-shot clips. For longer content, generate several clips and stitch them together.
The model runs on a GPU that starts on demand. The first launch can take up to a minute while it warms up; later generations are faster.
Whatever you generate is yours to use, including for commercial work, subject to the underlying open model's licence. You're responsible for your prompts.
Generate a starting frame with our free AI Image Generator, then animate it here.