Seedance 2.5 Edit & Extend on Segmind: change a clip or continue it
Two new ways to work with ByteDance Seed's Seedance 2.5 video model. Edit a clip from a text prompt — swap the subject, add or remove objects, or change the audio. Or continue any clip forward or backward with new footage that keeps its look, motion and audio.
Seedance 2.5 Edit and Seedance 2.5 Extend add two new jobs to ByteDance Seed's Seedance 2.5 video model. Seedance 2.5 Edit rewrites a video you have from a text prompt — swap the subject, add or remove an object, or change the audio. Seedance 2.5 Extend carries a clip forward or backward with new footage. Both read a source clip and a plain instruction, keep its aspect ratio, and carry its native audio. Both render at 480p, 720p or 1080p and take up to 30 reference images. Real human faces are declined, so use animated, stylised, product or landscape footage.
Edit
Seedance 2.5 Edit takes a source video and a prompt that names one change — add, remove, replace or change something, or edit the audio. Everything you do not mention stays as it was, so a subject swap keeps the original camera, motion, background and lighting. Add up to 30 reference images, cited as @Image 1, to hold a character or style. The output matches the source's shape and length, so there is no duration or aspect-ratio field to set. It suits swapping a product or character in approved footage, clearing an element a client wants gone, or replacing a soundtrack while the dialogue stays.
ModelSeedance 2.5 EditEdit video with a prompt: swap subjects, objects, audio.Replace the tin toy robot in @Video 1 with a soft plush teddy bear of the same size, keeping its exact marching motion, the workshop bench, the colourful building blocks and the lighting unchanged.
The tin robot becomes a plush teddy bear with the same marching gait, while the bench, blocks and lighting stay put; name what must stay so only the subject changes.
Add a small red wind-up toy train rolling across the front edge of the workshop bench in @Video 1, keeping the marching tin toy robot, the colourful building blocks and the lighting exactly as they are.
A red toy train rolls across the foreground while the robot keeps marching and the blocks stay in place; use 'add' to introduce something new and 'remove' or 'replace' to change it.
Restyle the marching tin toy robot in @Video 1 to match the ornate brass clockwork robot in @Image 1 — brass panels, exposed gears and a glass dome head — keeping its exact marching motion, the workshop bench, the colourful building blocks and the lighting unchanged.
With one reference image on @Image 1, the robot takes on the brass, riveted, gear-filled look while its stride and the room stay untouched; a clean single-subject reference carries a style further than adjectives.
Use edit from the API
Send the source video and your edit instruction. The response returns the edited clip at the source's shape and length.
# pip install "segmind>=1.1.0"
# export SEGMIND_API_KEY="YOUR_API_KEY"
import segmind
result = segmind.run(
"seedance-2.5-edit",
video="https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-edit-source-739a4cc6.mp4",
prompt="Replace the tin toy robot in @Video 1 with a soft plush teddy bear of the same size, keeping its exact marching motion, the workshop bench, the colourful building blocks and the lighting unchanged.",
resolution="720p",
generate_audio=True,
)
print(result.get("output"))# 1. Submit the request; the reply carries its request_id
REQUEST_ID=$(curl -s -X POST "https://api.segmind.com/v2/seedance-2.5-edit" \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"video": "https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-edit-source-739a4cc6.mp4",
"prompt": "Replace the tin toy robot in @Video 1 with a soft plush teddy bear of the same size, keeping its exact marching motion, the workshop bench, the colourful building blocks and the lighting unchanged.",
"resolution": "720p",
"generate_audio": true
}' | jq -r .request_id)
# 2. Check its status until it is COMPLETED
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID/status" \
-H "x-api-key: YOUR_API_KEY"
# 3. Fetch the result
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID" \
-H "x-api-key: YOUR_API_KEY"Extend
Seedance 2.5 Extend takes a source video, a direction and a prompt for the new section. Forward continues the clip after its last frame; backward adds what happened before its first. Set a duration of 4–30 seconds, or leave it on auto, and add reference images, videos or audio to keep a continued subject on-model. The output is the new footage only, so you join it to your original clip. It suits lengthening an approved shot that ran a beat short, or chaining shots by feeding the returned last frame into the next run.
ModelSeedance 2.5 ExtendExtend video forward or backward, keeping look and motion.The wind-up tin robot keeps marching to the end of the bench, reaches the row of colourful wooden building blocks and bumps the stack with its chest; the blocks topple and tumble across the workshop bench as the robot's legs keep ticking. Same stiff clockwork gait, same bench and window light, natural workshop ambience with the clatter of falling blocks.
The robot reaches the blocks and topples them while its gait, framing, window light and audio carry on without a seam; the output is the new footage only, so stitch it after your clip.
A small yellow tin wind-up toy duck, matching @Image 1, waddles into frame from the left as the tin robot keeps marching; the two toys meet beside the row of colourful wooden blocks and pause nose to nose on the sunlit workshop bench. Keep the robot's stiff clockwork gait, the same bench, window light and natural workshop ambience.
A yellow tin duck matching @Image 1 waddles in and meets the marching robot, so the new character stays on-model while the bench, light and robot carry over from the source.
Use extend from the API
Send the source video, a direction and a prompt for the new section. The response returns the added footage to join onto your clip.
# pip install "segmind>=1.1.0"
# export SEGMIND_API_KEY="YOUR_API_KEY"
import segmind
result = segmind.run(
"seedance-2.5-extend",
video="https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-edit-source-739a4cc6.mp4",
direction="forward",
prompt="The wind-up tin robot keeps marching to the end of the bench, reaches the row of colourful wooden building blocks and bumps the stack with its chest; the blocks topple and tumble across the workshop bench as the robot's legs keep ticking. Same stiff clockwork gait, same bench and window light, natural workshop ambience with the clatter of falling blocks.",
duration=6,
resolution="720p",
generate_audio=True,
)
print(result.get("output"))# 1. Submit the request; the reply carries its request_id
REQUEST_ID=$(curl -s -X POST "https://api.segmind.com/v2/seedance-2.5-extend" \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"video": "https://segmind-resources.s3.amazonaws.com/input/seedance-2.5-edit-source-739a4cc6.mp4",
"direction": "forward",
"prompt": "The wind-up tin robot keeps marching to the end of the bench, reaches the row of colourful wooden building blocks and bumps the stack with its chest; the blocks topple and tumble across the workshop bench as the robot'\''s legs keep ticking. Same stiff clockwork gait, same bench and window light, natural workshop ambience with the clatter of falling blocks.",
"duration": 6,
"resolution": "720p",
"generate_audio": true
}' | jq -r .request_id)
# 2. Check its status until it is COMPLETED
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID/status" \
-H "x-api-key: YOUR_API_KEY"
# 3. Fetch the result
curl -s "https://api.segmind.com/v2/requests/$REQUEST_ID" \
-H "x-api-key: YOUR_API_KEY"Shared settings
| Setting | Value | Members | When |
|---|---|---|---|
| resolution | 480p, 720p (default), 1080p | Both | 480p to draft, 720p for finals, 1080p (10-bit) for the highest fidelity |
| generate_audio | true (default) or false | Both | Leave on to keep or continue the soundtrack; off for a silent clip |
| reference_images | Up to 30, cited @Image 1 | Both | Hold a character or style across the edit or continuation |
| direction | forward or backward | Extend | forward continues after the last frame; backward adds a lead-in before the first |
| duration | 4–30 s, or -1 for auto | Extend | Sets the length of the new footage; Edit has no duration and inherits the source's length |
| reference_videos | Up to 9 extra clips, cited @Video 2 | Extend | Carry motion or style from other clips into the continuation |
Good to know
- Real human faces in the source are declined by the content policy, so use animated, stylised, product or landscape footage.
- Both jobs bill the source clip on top of the output, so an edit or extension costs more than a plain clip of the same length.
- Seedance 2.5 Edit has no duration or aspect-ratio field: the output always inherits the source's length and shape, and both are rejected if sent.
- Seedance 2.5 Extend returns the new footage only, not the original with the continuation joined on, so you stitch the two clips together yourself.
- Rendering is always at native resolution; to enlarge a result, pass it through the segmind-video-upscaler model.
Resources
- Try Seedance 2.5 Edit on Segmind · API reference
- Try Seedance 2.5 Extend on Segmind · API reference
- Seedance 2.5 - ByteDance Seed — the maker's overview: 30-second storytelling, precise reference control, editing and extension, green-screen and white-model control, and camera and performance blocking.
- Introducing Seedance 2.5 - ByteDance Seed — the official launch write-up: 30s single-pass generation with multi-round extension, multimodal referencing, and precise editing including green screen and camera perspective.
- #seedance-2-5
- #bytedance-seed
- #model-launch
- #video-to-video
- #video-editing
