Make video greetings with sound using Google Veo

Draft on Lite, send on Veo 3.1

The three share almost everything: the same lengths, two resolutions, wide 16:9 or tall 9:16, sound on or off. What changes is the price per second and the finish. Veo 3.1 is the full model and costs more per second than any other video model we run. Fast sits in the middle; Lite is the budget version.

The clips above show the split. Mum blowing out her candles as the family finishes the song, and the newlyweds clinking glasses by the sea at sunset, were made with Veo 3.1 at 1080p: faces close, light that matters, a moment someone will watch ten times. The rooftop toast is Veo 3.1 Fast at 720p; the door flying open onto confetti is Veo 3.1 Lite.

A sensible habit: run the prompt on Lite, fix what the draft got wrong, then spend the credits on Veo 3.1 for the version you send. Fast suits a clip that should look better than a draft but is a group-chat joke, not a keepsake.

Singing, cheering, waves: writing sound into a Veo prompt

Veo's sound is more than background. It does voices: a family singing the last line of Happy Birthday, friends shouting surprise, a short spoken toast. It does the room too - clapping, a chair scraping back, waves, traffic far below a rooftop. You get what you describe, so describe it.

Write the sound into the prompt plainly, the way you would tell a friend on the phone: sounds of waves, glasses and cheering. If someone speaks, put the words in quotes and keep them to one short sentence - eight seconds goes quickly.

Two honest limits. Veo chooses the voices itself, so the clip will not sound like Grandma. If the exact words in a voice you picked matter more than the scene, make a talking photo card with Kling AI Avatar v2 or OmniHuman 1.5; for a real song with verses there is Lyria. And if the clip will play muted in a group chat anyway, switch sound off and pay less per second.

Past 8 seconds or on a tight budget: where Veo hands over

Veo stops at 8 seconds. For a longer take - a whole toast, say - Kling 3.0 goes up to 15 seconds at 1080p with sound on or off, and MiniMax H3 and Wan 3.0 Prime go up to 10.

For value, MiniMax H3 is our default: Auto picks it for video, every video scene in our catalogue is animated with it, and its clips always come with sound. For a calm, steady loop, try Wan 3.0 Prime. For a square clip made from words alone, Kling 3.0, MiniMax H3 and Wan 3.0 Prime all do 1:1; Veo makes 16:9 and 9:16 only.

Veo earns its price when the moment is short and loud and the sound carries half of it: candles going out, glasses meeting, a door flying open. For a long, quiet moment - a grandfather on a bench, a slow dance - you will usually get more for your credits from Kling 3.0 or MiniMax H3.

Push-ins, candle smoke and a frame that fills a phone

We list Veo 3.1 under Cinema quality, and it is most at home in warm low light: candle smoke, a lamp over the kitchen table, the last sun on a terrace. It follows the camera move you write into the prompt, so name one: a slow push-in, a dolly along the table, an orbit around the couple. The candles clip above asked for a slow push-in in a cosy kitchen and nothing more.

Every clip on this page started from a scene picture in our catalogue, drawn with Nano Banana Pro. Yours can start from a photo of your own people or from a card you made here with the whole family in it. Then think about shape as much as light. Veo makes tall 9:16 clips, the shape of a phone held upright, and wide 16:9 ones for a laptop or the TV at the party. A vertical birthday clip fills the screen in a messenger instead of sitting between black bars.