Extending Videos — References (v1.11)

In the References section, click the Source Video slot to load the original video (.mp4 or .mov), or drag it directly from the History panel on the left.


Hover over the video thumbnail to view an enlarged preview.

After the original video is loaded in, click a Reference slot next to it to add inputs that guide video editing and improve consistency in actors or environments. Supported reference types include AI Actors, images, videos, and audio.


Before adding references, check the requirements for each media type, including format, aspect ratio, resolution, file size, and duration, as shown below. If a reference does not meet the requirements, a compatibility warning dialog will identify the issue preventing the upload.

AI Model	Seedance 2.0 & 2.0 Mini (Maximum 15 References)				Seedance 2.5 (Maximum 50 References)
Requirements	AI Actor	Image	Video	Audio	AI Actor	Image	Video	Audio
Max number	Up to 7	Up to 9 (Including AI Actor)	Up to 3	Up to 3	Up to 10	Up to 30 (Including AI Actor)	Up to 10	Up to 10
Formats	iModel	PNG, JPG, JPEG, WebP	MP4, MOV	MP3, WAV	iModel	PNG, JPG, JPEG, WebP, BMP, GIF	MP4, MOV	MP3, WAV
Aspect ratio	--	0.4–2.5	0.4–2.5	--	--	0.4–2.5	0.4–2.5	--
Width / Height		300–6,000 px	300–6,000 px	--		300–6,000 px	300–6,000 px	--
File size		30 MB max per image 64 MB max total	50 MB max per video 200 MB max per 4K video	15 MB max per audio 64 MB max total		30 MB max per image	200 MB max per video	15 MB max per audio
Duration		--	2–15 sec	2–15 sec		--	4–30 sec	2–30 sec

AI Actors and images share a combined limit. For example, you can use 5 AI Actors and 4 images, or 2 AI Actors and 7 images. If a character is fighting zombies in the original video, you can add the character’s AI Actor, introduce additional characters in subsequent shots, and add reference images of objects to continue the story.

• AI Actor

  After creating custom AI Actors, you can apply them from the AI Actor Library to maintain consistent character appearances in generated videos. Click the Reference slot and select AI Actor. Choose one or more custom actors from the AI Actor Library, then click Apply AI Actors.
  
• Image

  Click the Reference slot and select Image to upload an image reference for an object, such as a vehicle, to include in subsequent shots. Alternatively, drag an image from the History panel on the right or Windows File Explorer.
  
• Video

  Select Video and upload a video reference to replace the original video's character motion, voice, visual style, composition, or background music.
• Audio

  Select Audio and upload audio to add or replace the original video's character dialogue or background music.

After adding references, use the Reference tools to filter reference types and check the current input counts and media durations.

1. Reference Filter: Filters references by All, Image, Video, or Audio, and displays the input count for each type.
2. Video Duration: Displays the current video input duration and the maximum duration supported by the selected AI model.
3. Audio Duration: Displays the current audio input duration and the maximum duration supported by the selected AI model.
4. Reset All:Removes all reference inputs.

Hover over a Reference thumbnail to preview the actor, image, video, or audio.

Each reference is represented by a label tag in the Prompt field for quick editing and reference assignment. Click the Reset Prompt button to clear the prompt field and start over if needed.

• Drag the tag to reposition it within the prompt.
• Click Copy on a reference thumbnail to insert its tag into the prompt. Alternatively, click Add Reference @ or type @ to select a reference tag from the context menu. Use the Image, Video, Audio, and AI Actor filters to show only references of the selected media type.
  

Enter a prompt describing how each reference should be used. You can enter a custom prompt and use a Large Language Model (LLM), such as ChatGPT, to refine it for better video generation with the selected AI model, or apply preset Camera Techniques to animate the camera.

Assign Video: Insert the [Source Video] tag into the prompt to connect the new shots to the End Frame of the original video and retain specific elements from the reference video. Assign AI Actors: Insert AI Actor tags, such as [Xiang Hua_3D] and [Amber_3D], into the prompt to maintain consistent character appearances across shots. Camera Control: Specify how the camera captures the character's movement. Assign Audio: Use an audio reference to start a new audio track, or have AI generate new audio based on the story instead of using the source video's audio. + Multi-Shot: Click the button to expand the shot prompt fields, such as Shot 1 and Shot 2. In the Main prompt, define shared elements such as the overall story, characters, and style. In each Shot prompt, describe the shot-specific character motions, composition, camera movements, and references. Click + Add Shot to add up to 10 shot prompt fields. Assign Image: Insert an image tag, such as [Image 1], into the prompt to add an object required by the plot.

Drag the bottom-right corner to expand the text field for longer descriptions.

For example, "Extending Source Video, maintain the appearance of the character Xiang Hua_3D. Change the appearance of Amber_3D so that she is covered in blood and dirt and has messy hair .

Use a follow shot to track the character's movement.
Start a new background audio track rather than continuing the audio from Source Video.

Shot 1: After Xiang Hua_3D shoots down a zombie, she nimbly vaults over the low wall while keeping her gun raised and aimed forward. She then runs in the same direction she is aiming to meet her friend. In the far background, the zombie she shot in Source Video remains lying on the ground.

Shot 2: Xiang Hua_3D hears her friend Amber_3D, calling from the rear of the passing pickup truck shown in Image 1. She then jumps onto the rear of the pickup truck and joins her friend.".

Set the video output options in the Render section, including resolution, video length, and audio settings.

• Ratio: Automatically maintains the original aspect ratio of the source video.
• Resolution: Select the desired video resolution from the options supported by the selected AI model. The output is generated at the selected resolution while automatically maintaining the source video's aspect ratio.
  
• Seconds: Select the video length from the Seconds drop-down list.
  
• Generate with Audio: Choose whether to include prompt-generated audio in the video. Enable this option when the prompt includes dialogue. Note that adding audio may incur additional charges.

Each submitted task uses available AI Points, Bonus Points, or DA Points. View your available points in the bottom-left corner of AI Studio, or click the ... button and select Account Settings to check your point balance.


To obtain more credits, click Plans to subscribe to an AI Service Plan for additional AI Points or purchase DA Points to top up your account.