Reference → video
Open the full StudioReference to video. Hold the look.
Add up to four pictures of a person, a product or a place, then write the scene. The model keeps what is in them across the clip, and the button shows the credits first.
Input Up to 4 picturesOutput VideoSeedance 2.0 Mini 45 credits
SampleAdd your photo first: Generate unlocks once it is in.
Video needs a plan or a credit pack.Open the full Studio
01 · Style references
Looks that hold.
Style references from our design library, labeled as such. They show the idea of a consistent world, not clips made on this page.
02 · How it works
Refs. Prompt. Roll.
For campaigns and series where the same face or product has to come back in every clip.
Add your references
Up to four pictures of who or what to keep: a face, an outfit, a product, a setting. JPG, PNG or WebP, up to 10 MB each.
Write the scene
Say what happens and how the camera moves. The references decide the look; the words decide the action.
Generate & keep
A 10-second clip, from 45 to 125 credits. It lands under the box and in your History; failed runs are refunded.
What it is
What is reference to video on Eidora?
Reference to video makes a clip from pictures you choose to keep and a prompt that says what happens. Unlike image to video, your pictures are not the first frame. They are references: the model studies the face, the clothes, the product or the place in them and puts those into a new scene. That is what you need when the same character or product has to appear in many different shots.
Reference to video on Eidora runs in the box at the top of this page and in the full video Studio. 2 models run this mode today, both Seedance models: Seedance 2.0 Mini and the sharper Seedance 2.0 Fast. Each takes up to four pictures and makes a 10-second clip without sound, priced from 45 to 125 credits.
Video needs a paid plan or a credit pack. A new account gets 6 credits at sign-up for still images, which is a good way to prepare clean reference pictures first. The price of every clip is on the Generate button before you press it; Pricing lists the plans and packs.
If you only need one picture to move, image to video is simpler and offers more models. If you have no pictures at all, start with text to video, or create the character with text to image and use those pictures as references here.
03 · Use cases
Where reference to video earns.
Campaigns with one face
Keep the same model, mascot or presenter across a set of ads, each clip in a new place or doing something new.
Product series
The same bottle, shoe or bag in a kitchen, on a beach and in a studio, without reshooting it for every scene.
Characters and stories
A character from your illustrations or earlier images that stays recognisable from one scene of the story to the next.
Outfits and styling
Put a garment from a flat-lay photo on a person in motion, or show one outfit in several settings.
04 · Models
Models for reference-guided video.
The two models that take reference pictures today. The prices are the ones the Generate button charges.
Video · 10 s
Seedance 2.0 Mini
Keeps your people and things, 10 s, no sound
45 credits · Plan or packVideo · 10 s
Seedance 2.0 Fast
Sharper, keeps your people and things, 10 s, no sound
125 credits · Plan or packAll 2 models in this mode
- Seedance 2.0 Mini45 credits
- Seedance 2.0 Fast125 credits
Guide
Holding a look across AI video: references, credits, privacy
Prepare references the model can read
The references do most of the work, so they need to be clear. For a person, use a sharp, front-facing photo with the face well lit, and add a full-length picture if the clothes matter. For a product, a picture on a plain background with no other objects is best. For a place, one wide picture that shows its character is enough.
Up to four pictures fit in one run, but more is not always better. Each picture should add something the others do not show. Pictures that disagree, for example the same person with different hair, make the model average them. If your pictures are low resolution, the AI image enhancer can sharpen them first.
Write the scene, not the subject
The prompt's job is the action and the camera. Point at the subject with a short name and let the pictures supply the details: “The woman in the red coat walks through a rainy night market, handheld camera following behind her, neon reflections on the wet ground.” Describing the face or the product again in detail can fight the references.
For a series, write each clip as its own scene and keep the references the same. Change only the place, the action and the camera from one clip to the next. Clips are 10 seconds without sound, so add music and voice when you edit them together.
Models, prices and refunds
Seedance 2.0 Mini is the lower-priced choice and the place to start. Seedance 2.0 Fast is sharper and costs more. Both keep your people and things, both make 10-second clips, and the price is fixed per clip: from 45 to 125 credits in this mode. See the Seedance 2.0 Mini page for the family's samples.
Credits come from a plan or a credit pack, and a run that fails is refunded automatically. More models will appear here as they pass our checks. Plans and packs are on Pricing, and billing questions are answered in Help.
Privacy and your reference pictures
Reference pictures often show real people, so treat them with care: use pictures you have the right to use. Pictures you upload are deleted automatically 30 days after upload and are never used to train AI models; the Privacy policy has the details.
Every clip is saved to your History, where you can watch and download it. When you need other modes, such as first and last frame, they are all in the video Studio.
05 · Questions
Mode questions. Straight answers.
How is reference to video different from image to video?
In image to video your picture is the first frame of the clip. In reference to video your pictures are references: the model keeps the people and things in them and puts them into a new scene you describe.
How many reference pictures can I add?
Up to four per run, JPG, PNG or WebP, up to 10 MB each. Two or three clear pictures usually work better than four mixed ones.
Which models support reference to video?
Reference to video on Eidora runs on Seedance 2.0 Mini and Seedance 2.0 Fast today. Both make 10-second clips without sound; Fast is sharper and costs more.
How many credits does a clip cost?
Reference to video on Eidora costs 45 to 125 credits per clip, depending on the model. The price is on the Generate button before you press it, and a run that fails is refunded automatically. Video needs a plan or a credit pack; see Pricing.
Can I use pictures of real people?
Only pictures you have the right to use, such as yourself, your products or people who agreed. Uploaded pictures are deleted after 30 days and are never used to train AI models.
Where do my clips go?
Every clip is saved to your History, where you can watch and download it. Keep the same references to make the next clip in the series.
Explore
Keep creating. Same studio.
Ready
Lock the look. Move it.
The composer is at the top of this page. The price is on the button, and failed runs are refunded.