Veo 3 in Telegram: video from a prompt

You have an idea for a clip but no camera, actors or editing skills. The bot offers another path: describe a scene in words and a neural network shoots it for you. 🎬

The bot's description mentions two Google tools for making video. Gemini Omni does multimodal generation from text and photos, from 89 coins. Veo 3 produces 4K or HD clips with realistic physics, lighting and sound.

Launch: one command

Everything starts with /start. Per the command description, you can send a voice message, text or a photo, and the bot turns it into a clip.

1. Press /start.
2. Pick a tool: Gemini Omni or Veo 3.
3. Describe a scene or attach a photo.
4. Choose the clip format.
5. Wait for the result. According to the bot's description, it takes minutes.

Two frame formats

Export comes in 9:16 and 16:9. The first fits vertical phone videos, the second suits horizontal viewing and regular video platforms.

Decide before generating: once coins are spent you cannot fix the format, and the clip has to be created again.

What is known about Veo 3 itself

Veo is Google's family of video generation models with sound. According to the documentation, Veo 3 supports clips of 4, 6 or 8 seconds, vertical and horizontal formats, and 720p and 1080p resolution. The next version, Veo 3.1 from October 2025, makes 8-second clips up to 4K with dialogue, effects and ambient sound.

So a one-minute video from a single request is not possible. These are short clips that you can join later in an editing program. 🎞️

How to write a prompt

• Who and where. Subject, place, time of day.
• Action. What happens in a few seconds.
• Camera. Close-up, orbit, static shot.
• Light and mood. Sunset, neon, an overcast day.
• Sound. City noise, music, a line of dialogue.

A short, concrete prompt works better than a long, vague one. If you do not like the result, change one detail at a time instead of rewriting everything.

Coins and costs

A good workflow: first state the idea in one sentence, then add two or three details and only then run it. That way you spend coins on a thought-out request, not a blind try.

The bot's services are paid with internal coins, and the cheapest Gemini Omni option starts at 89 coins. Check package prices and the cost of individual generations in the bot before paying.

Neural networks do not always nail it on the first try. Budget for several attempts per clip and do not spend everything on the first generation.

Rights and honesty

Do not create clips of real people without their consent, and do not pass generated video off as documentary footage. Follow the rules of the platform you upload to: many require labeling AI content.

Examples of work and prompt tips are published in the project's channel.