Three habits that make image-to-video results more predictable.
Settle composition, colour and character in Nano Banana first, where reruns are cheap. Download the frame you like and upload it here as the starting image.
"She turns toward the window" is subject motion; "slow push-in" is camera motion. Naming both in the AI video generator prompt keeps the shot from drifting.
Some versions add sound, some accept a last frame, some run longer. The settings panel shows what the selected model supports before you spend credits.
The AI video generator holds seventeen versions from Seedance, Veo, Kling, Wan, Grok, Hailuo and HappyHorse. It opens on Seedance 2.0, which needs a Basic plan; versions open on the free tier are marked in the version bar. Every run needs a prompt. Many versions also take a first frame, and some take a last frame, reference images, reference video or audio. To animate a still made on this site, download it from My Creations and upload it as the first frame; there is no automatic hand-off. Choose duration, resolution and aspect ratio where the model offers them, generate, and the clip is saved to My Creations.
Each tab is a different way to start a clip.
Describe the scene, the action and the camera. Useful when you have no image yet and want to explore movement ideas.
Scene in Words
Subject, setting and action.
Camera Directions
Pan, tilt, dolly or static.
Duration by Model
Short to longer clips.
Six capabilities across the video models.
Seven video model families.
Start from any still you upload.
Generated audio where supported.
Durations vary by model.
Resolution options per model.
Every video kept in My Creations.
Three groups, two workflows each.
Generate the cover scene as a still, upload it to the AI video generator as the first frame and ask for subtle motion such as drifting light or steam for an audiogram background.
Pick a vertical format, describe a short visual hook for the episode topic and pair it with your audio in an editor.
Design the invitation scene as a still, then add gentle motion such as falling petals or a slow camera move for a shareable clip.
Upload a venue photo and describe the decoration and lighting moving into place, so clients see the mood before the day.
Use key art as the first frame and describe the camera move. Treat what the AI video generator returns as a pitch or storyboard shot, not final footage.
Animate a still from your game's art style into a short loop for a store page or social post, checking that characters stay on model.
Three actions on this page.
Upload a still as the first frame in the AI video generator, or start from text.
Choose a version, then duration, resolution and sound where offered.
Watch the clip in My Creations and download it.
Inputs, lengths, sound and cost.
Yes. Download the image from My Creations and upload it as the first frame here. There is no automatic transfer between the image and video pages.
Seventeen versions from Seedance, Veo, Kling, Wan, Grok, Hailuo and HappyHorse. Each brings its own durations, resolutions and inputs.
It depends on the model: for example 5, 10 or 15 seconds on Seedance 2.0 and Kling 3.0, 4 to 12 on Seedance 1.5 Pro, and 6 or 10 on Hailuo 2.3.
Only on models that offer it, such as Seedance 2.0, Seedance 1.5 Pro and Kling with the sound option on. Others return silent video.
Some versions are text-only or use a separate mode for images. Switch to a version with a first-frame or image-to-video mode, and the upload slot appears.
The default Seedance 2.0 needs a Basic plan, but several versions are marked Free in the version bar. New accounts get 10 credits once, valid for 30 days, which may not cover longer or higher-resolution clips.
Image pages that feed the AI video generator.
An AI video generator for people who already have the picture: start from a Nano Banana still, upload it as the first frame, and describe how the subject and the camera should move. Seventeen video model versions, clips of roughly 4 to 15 seconds depending on the model.
0 / 20000
