Semilla 2.0

Seedance 2.0 features exclusive optimizations for image-to-video and text-to-video scenarios with its intelligent lens switching function. No professional editing knowledge or complex lens commands are needed—simply with an image sequence or text description, the model automatically matches the suitable lens switching methods. It supports a variety of transition effects such as fade in/out, pan, tilt, push, pull, and seamless scene transitions, and can adjust the switching speed based on frame content and plot rhythm, effectively boosting transition naturalness by 92%. This turns static images or text-described scenes into dynamic videos with naturally flowing camera movements, directly generating a complete video with smooth lens switching without requiring any additional description of transition details.

This is a practical and efficient core feature of Seedance 2.0. When you’re unsatisfied with a small detail in a generated video, there’s no need to start over from scratch—it supports targeted revisions including character replacement, clip deletion, and movement adjustment. You can modify specific parts accurately without altering the overall frame, rhythm, or style of the original video, eliminating the trouble of remaking the entire video.
If the lead singer in a band performance video doesn’t meet your expectations, upload a new image of the lead singer and write the prompt: "Replace the female lead singer in @Video 1 with the male lead singer in @Image 1, replicating all original movements exactly". A new video is generated quickly, reducing revision time by 70%.

The most frustrating creative pain points—character face distortion, lost product details, abrupt scene changes, inconsistent lens styles—are completely solved in Seedance 2.0. The model maintains high consistency throughout the video, from facial features and costumes of characters to small text and prop details in the frame, greatly enhancing the video’s integrity and professionalism.
When creating a plot short video, from the opening corridor scene to the warm indoor ending, the character’s facial features and costume style remain unchanged throughout, improving frame consistency by 95% and delivering a smoother viewing experience for the audience.

Seedance 2.0 not only generates high-quality frames but also achieves accurate sound timbre reproduction and natural sound effect matching. It supports multi-language line generation (e.g., Spanish, English), and details such as equipment collision and environmental sound effects are presented synchronously. No additional audio editing is needed—the generated video comes with an immersive audio-visual experience out of the box.
For a tactical short film, make characters deliver precise Spanish lines paired with sound effects like equipment collisions and breathing, boosting audio-visual matching by 85% to create a cinema-level auditory experience.

This is the standout core feature of Seedance 2.0. It not only accurately replicates movements, cinematography, special effects, and sound timbres from reference materials, but also enables smooth extension and connection of existing videos to achieve a "continue shooting" effect. Whether it’s a Hitchcock zoom from a movie or the beat-matching rhythm of a viral video, just upload the reference video and the model replicates it 1:1—no more lengthy prompts required.
If your plot short film is missing only the ending, upload the original video and write the prompt: "Extend @Video 1 by 15 seconds. Refer to the character riding a motorcycle from @Image 1 and @Image 2, and add a creative ad segment". A naturally connected ending is generated instantly, enhancing video extension naturalness by 90%.

Seedance 2.0 supports mixed input of up to 9 images, 3 videos (total duration ≤15s), 3 audio files (MP3 format / total duration ≤15s) plus natural language text, with a maximum total of 12 files for mixed input. You can set the visual style with a single image, specify character movements and cinematography with a video, set the rhythmic mood with a few seconds of audio, and add details with text. It’s just like a director arranging on-set materials, ensuring every creative idea is accurately realized.
For a product promotion short film, upload real product photos + a reference advertising cinematography video + brand background music, then write the prompt: "Close-up of product details + slow lens push-in". No from-scratch editing is needed—a tailored short film is generated in 1 minute, improving creative efficiency by 80%.