AI video
Describe a scene. Get a video with sound.
Describe the scene, the action, the camera and the sound in a GNJ AI chat, or send a photo to use as the first frame, and get a 4–10 second video with sound, in landscape or vertical. Ask for changes in your own words, or trim, join and caption clips. Powered by Google’s Veo 3.1 and Omni video models, and credits are used only when a video is delivered.
Original demonstration
TIDE · Morning surf
Open TIDE to watch an 8-second video made from one request: “a golden retriever running through the shallow waves at dawn, low tracking shot”. Turn the sound on to hear the audio generated with the picture.
From brief to result
- 01
Describe the shot: the subject, the action, the setting, how the camera moves and what we hear. Or start from a photo as the first frame.
- 02
Say landscape (16:9) or vertical (9:16) and how many seconds, or keep the defaults.
- 03
Most videos are ready in 1–3 minutes and play right in the chat. If one takes longer, it’s posted to the conversation when it’s done.
- 04
Ask for changes in your own words, like “make it snow” or “turn it into a watercolor”. Or have it trim, join or caption your clips.
What you can make
- Sound included: ambience, effects and dialogue are generated with the picture.
- Bring photos to life: use your photo as the first frame, and optionally choose the last frame too.
- Landscape or vertical: 9:16 is ready for TikTok, Reels and Shorts.
- Keep refining: change the style, the weather or what’s in the scene while keeping the original motion.
- Credits are used only for delivered videos, never for failures or refusals. An 8-second 720p video is about 107 credits.
Make an 8-second vertical video: a city rooftop at dusk, an iced coffee on the railing with condensation sliding down the glass, lights coming on across the skyline as the camera slowly pushes in. Sound: wind and distant traffic.
What to bring
- The picture: who or what is in it, what happens, where, and in what light.
- Camera and sound: push in, follow or orbit; which ambience, effects or lines of dialogue you want.
- A photo (optional): send one to start from, or to end on, a particular frame.
Review and refine the result
- Check the action and the camera first, then ask for one or two specific changes at a time.
- For something longer, make several clips and have the AI join them in order and add captions.
- Before you publish, check that no unexpected people, logos or text appear in the frame.
Scope and requirements
- Each clip is 4–10 seconds (10 seconds with Omni Flash only); make several clips and join them for something longer.
- Videos are 720p; Omni Flash can also make lower-cost 360p clips.
- Real public figures can’t be generated. Get permission before using a photo of someone else.
FAQ
A few good questions
How much does a video cost?
Videos use credits, and only delivered videos are charged; failures and refusals are free. With the default Omni Flash model, an 8-second 720p video is about 107 credits, 10 seconds is about 133, and 360p costs less. Editing a clip costs twice the clip’s price. Current prices are shown in the app.
Which models make the videos?
Google’s Veo 3.1 and Omni video models, with Omni Flash by default. Ask for Veo 3.1 Lite, Veo 3.1 Fast or the highest-quality Veo 3.1 in the chat when you need them.
Can I use my own photo?
Yes. Send the photo in the chat and describe how the scene should move; the AI uses it as the first frame, and you can add another as the last frame. Real public figures can’t be generated.
Can I edit the result?
Yes. Describe the change and the AI regenerates the clip while keeping its motion. Trimming, joining clips, changing the speed and adding captions also happen right in the chat.
Start your next task with a conversation.
Bring a question, a file, or an idea that’s not quite there yet.