Text-to-video basics

The video generator turns a prompt — or an image — into a moving clip. It has three tabs: Create Video, Edit Video, and Motion Control. Most work starts on Create.

One generator, four input modes

There's no mode switch — PonPon infers what you want from what you give it:

• Text-to-video — a prompt and nothing else. Maximum freedom; the model invents every frame.
• Image-to-video — upload a Start Frame and the model animates it as the first frame. Maximum control over the look.
• Start → end morph — add an End Frame too, and the clip transitions from one image to the other.
• Reference-to-video — attach reference images/videos on a capable model to carry a subject or style into the shot.

Write motion, not just a scene

A still-image prompt describes a moment. A video prompt describes a moment that changes — subject, action, camera, and pacing:

A surfer paddles out and stands up on a wave at sunrise, camera tracking alongside at water level, spray catching the light. Smooth, cinematic motion.

Choosing a model

The picker is a row of chips. Each has a clear strength:

• Veo 3.1 — the most controllable camera language plus native audio. A great all-rounder. Veo 3.1 Fast drafts the same look quicker.
• Sora 2 — best-in-class physics and texture realism, with synced audio.
• Kling 3.0 — precise motion, lip-sync, and multi-shot storytelling (several camera cuts in one generation).
• Seedance 2.0 — fast and expressive, vertical-first, with audio-visual beat sync. Seedance 2.0 Fast is faster still.
• HappyHorse — the most versatile: text, image, reference, and editing pipelines, with many reference characters and native audio.

Aspect ratio, duration, resolution, audio

• Aspect ratio — 16:9 for YouTube, 9:16 for TikTok / Reels / Shorts, 1:1 for feed (hidden when you start from an image).
• Duration & resolution — the options depend on the model.
• Audio — for audio-capable models a toggle generates sound with the picture; some models (like HappyHorse) always include it.

Beyond Create: Edit and Motion Control

• Edit Video — feed in an existing clip and a prompt to restyle or modify it (video-to-video), optionally keeping the original audio.
• Motion Control — drive a still character image with the motion from a reference video, choosing whether the character follows the image or the video.

After the render

• Sequence shots and re-run them in Flow, or build a multi-scene piece in Studio.
• Add a voiceover, music, or sound effects in the audio studio.

For the deeper method — camera language, shot structure, and common fixes — read Prompting for video.