Character you already have
Start from a photo or a written description and keep the same face, hair and outfit across every clip, instead of generating a new stranger each time.
Write a scene or upload a character photo and the generator animates her: a glance over a coffee cup, a walk through a bookshop, a smile at the window. Pick a ratio, choose a model and get a 5–10 second clip you can post, edit into a story or use as a character test.
Generate Video









Prompt Guide
Model
Reference Image
Scenes generated from a photo plus a prompt — every card shows the prompt that produced it.
A woman sitting by a café window glancing up with a gentle smile, warm bokeh lights in the background, soft cinematic lighting, slight head movement, 5 seconds, 16:9.
Try this promptShe walks through a second-hand bookshop, fingertips brushing the spines, warm lamp light, slow tracking shot, natural movement.
Try this promptMorning routine in a sunlit kitchen, she lifts a mug and smiles at someone off-camera, soft window light, medium shot.
Try this promptShe walks through a green park on a bright day, hair moving in the breeze, camera follows at walking pace, shallow depth of field.
Try this promptStudio portrait, she slowly turns her head to camera, rim light on her hair, dark neutral backdrop, 5 seconds.
Try this promptFashion portrait, silk fabric catching soft window light as she shifts her weight, calm expression, subtle camera push in.
Try this promptEvening city street, she turns in a red dress, warm streetlights behind her, gentle camera move, 5 seconds.
Try this promptWhat this page is built to do.
Start from a photo or a written description and keep the same face, hair and outfit across every clip, instead of generating a new stranger each time.
The prompt controls the moment — a glance, a half-smile, a laugh — so clips read as acting beats rather than a slow zoom on a still image.
No photo yet? Describe her and the scene. Have one? Upload it and add the action you want.
9:16 for shorts and stories, 16:9 for YouTube, 4:3 and 1:1 for posts — chosen before generation, not cropped afterwards.
5 and 10 second outputs stay in the same style, so three clips can be cut into one character scene.
Compare Grok Imagine Video, Seedance 2.0 and Kling 3 on the same prompt without moving your project.
Three steps from idea to final render.
Write who she is, what she does and how the camera behaves — or upload a character photo and describe only the action.
9:16 for social, 16:9 for YouTube; 5 seconds for a beat, 10 seconds for a fuller action; keep the free model for the first try.
Check the face, hands and motion, adjust one detail in the prompt, then download the clip for editing or posting.
Same prompt, different motion and likeness.
| Criteria | Grok Imagine Video 1.5 | Seedance 2.0 |
|---|---|---|
| Best role here | Free first pass to test the scene and the character. | Final clip when the motion and likeness must hold. |
| Pick it when | You are exploring framing or a new photo. | You are publishing or cutting several clips together. |
| Free trial | One free generation per new account. | Paid only. |
Illustrative scenarios with fictional names.
I used one photo of my character and made three clips in the same light. They cut together like a tiny scene instead of three unrelated shots.
Writing the beat — glance up, then smile — was the trick. The first clip felt like a photo with a zoom until I described the action.
The 9:16 output went straight into shorts without cropping. That alone saved me an editing pass.
It turns a written scene or a character photo into a short video clip of that character — expression, movement and camera framing included — without filming or 3D animation.
You choose 5 or 10 seconds per generation. Clips share a consistent look, so several can be cut into one longer scene in any editor.
No. You can describe the character and the scene from scratch. Uploading a photo helps when you need the same face across multiple clips.
Character tests, short films, social posts, storyboards and mood boards. Check the model terms before commercial use.
The first generation on a new account runs on the free entry model. Additional models such as Seedance 2.0 and Kling 3 are paid.