BytePlus Seedance 2.5 multimodal video generation, the next generation after Seedance 2.0 with long-form storytelling, omni references, and controllable editing. Inputs are greatly expanded: up to 50 reference assets per request (up to 30 images, 10 reference videos, and 10 audio clips). Freely combine text, images, videos, and audio to generate a video, edit a reference video (add or remove objects, replace the subject), or extend it forward or backward. The control panel offers an aspect-ratio selector with 7 options: 16:9 (landscape, YouTube), 9:16 (portrait, IG Reels / TikTok), 1:1 (square, IG posts), 4:3, 3:4, 21:9 (ultrawide cinematic), and adaptive (follows the reference assets); choose the resolution (720P by default, or 1080P) and the video length (4 to 30 seconds, 15 by default). Upload a first-frame image to set the opening shot and a last-frame image to set the ending shot; upload multiple keyframe images to chain scenes in order. When a first/last frame is uploaded, the aspect ratio automatically follows the image. Refer to assets in the prompt as @Image 1, @Video 1, and @Audio 1 to assign their roles. Dialogue generation natively supports more than 10 languages including Chinese, English, Japanese, and Korean - great for product ads, story shorts, music videos, and multilingual marketing clips. The result downloads as an MP4 file. Usage: 1. Prompt only, no file uploads. 2. Prompt + first-frame image (optionally + last-frame image). 3. Prompt + keyframe images. 4. Prompt + reference images / videos / audio clips (assign roles with @Image 1, @Video 1, @Audio 1). First-frame or last-frame images cannot be uploaded together with keyframes or reference images, videos or audio (a first frame may be paired with a last frame). Uploaded images (first frame, last frame, keyframes, reference images) that contain real people are declined by the AI service under its privacy policy; use illustrations, products, scenes or AI-generated characters instead. Generated videos follow the control-panel length (4 to 30 seconds, 15 by default) and resolution (720P by default, or 1080P); the aspect ratio follows the control-panel selection (default 16:9). With a first/last frame or a reference video the aspect ratio follows the asset automatically, and with a reference video the model decides the video length (up to 30 seconds). 1080P videos use 10-bit H.265 encoding, which some browsers or devices cannot play directly; download the file and use a player that supports H.265. Maximum total upload size per request: 512MB (all files and message content combined). Accepted image formats: bmp, gif, heic, heif, jpeg, jpg, png, tif, tiff, webp. Maximum number of images: 30. Maximum size per image: 30MB. Image dimensions: 300 to 6,000 pixels per side. Accepted video formats: mov, mp4. Maximum number of videos: 10. Maximum size per video: 200MB. Video duration: 2 to 30 seconds (total at most 30 seconds). Video dimensions: 300 to 6,000 pixels per side. Video resolution: 407,696 to 8,295,044 pixels in total. Video aspect ratio: 0.4 to 2.5. Video frame rate: 24 to 60 fps. Accepted audio formats: mp3, wav. Maximum number of audio files: 10. Maximum size per audio file: 15MB. Audio duration: 2 to 30 seconds (total at most 30 seconds). Output content formats: Video Pricing: High Video output: one MP4 video of 4 to 30 seconds (15 by default) at 720P or 1080P, with the length, resolution and aspect ratio following the panel selection (with a reference video the model decides the aspect ratio and length automatically, up to 30 seconds). This module does not support continuous conversations. AI can make mistakes. Please verify important information.