Create a character or upload photos of one, then describe the next scene. ClipNova builds every clip from your reference images, so the face, hair and outfit stay the same from scene to scene, in 9:16, 16:9 or 1:1.
Here is one fictional character, Iris, followed from her first portrait to a run of scenes. Every picture below was made for this page.
Start in AI Characters: describe who your character is and pick one of 8 looks. Each run makes one image, so generate until one feels right. Already have a character? Upload your own photos or artwork instead.
Open Image to Video and switch to the Images to Video mode. Add your character as references: a portrait, a full body shot and a side angle work best. Base takes 1 to 7 images, Pro 2 to 4 and Ultra 1 to 9. Clear, well lit photos of one outfit give the steadiest results.
Write the scene like a shot: the place, the action, the light and the camera move. ClipNova renders a clip that keeps your character's face, hair and clothes from the references. Single clips run 4 to 10 seconds depending on the tier.
For the next scene, keep the same reference set and change only the prompt: a rainy street at night, a rooftop, a train. Every clip starts from the same pictures, so the character reads as one person across the whole story. Ultra also renders native sound.
Need more than one clip? Set a longer length, up to 5 minutes, and ClipNova plans the video shot by shot with your references in every shot. Add an AI voiceover, music and captions, or download single clips and cut them together yourself.
Stories, cartoons, mascots and creator channels all fall apart when the main character changes face between clips. Each example below is fictional and was made for this page.
A prompt only video model invents a new person every time you generate. Change one word and the hair, the face or the jacket changes with it. That is fine for one clip and a problem for a story, an ad campaign or a channel built around one character.
ClipNova's Images to Video mode works the other way round: you show the model who the character is with up to 9 reference images, and every clip is built from them. The order does not matter and they are not keyframes; they are the identity each shot has to keep. Base uses Vidu Q2, Pro uses Kling 1.6 Elements and Ultra uses Seedance 2.5, which also renders native sound.
What it does not do: there is no training step and no custom model, and it cannot fix weak references, so use clear, well lit photos of one consistent outfit. Only use photos of real people who have given you permission.
Only have a description? Start with the AI Character Generator to create the face. For products, places or several subjects in one clip, see Reference to Video AI.
Planning a whole story? Our list of the best AI video generators for storytelling compares tools for narrated and character led videos.
A tool that keeps the same character, with the same face, hair and outfit, across many separate scenes. In ClipNova you give the Images to Video mode a set of reference images of your character and describe each new scene; every clip is built from those references.
Base takes 1 to 7 images, Pro 2 to 4 and Ultra 1 to 9. A portrait, a full body shot and a side angle of the same outfit usually give steadier results than a single photo.
Yes. References can be photos, drawings or 3D renders. Keep one style across your reference images and describe each scene in that same style.
Single clips run 4, 6 or 8 seconds on Base and Ultra and 5 or 10 seconds on Pro. Ask for a longer length and ClipNova plans the video shot by shot, up to 5 minutes, with your references in every shot. Formats are 9:16, 16:9 and 1:1.
Only yourself or someone who has given you permission. Do not upload photos of celebrities or other people without consent, and label realistic AI people where the platform asks for it.
Plans start at $19/mo. Each video uses credits from your monthly allowance based on tier and length, and the exact cost is shown before you generate.
Join creators turning ideas into scroll-stopping videos, no crew, no software, no learning curve.
