Video Extend
Add 3, 5 or 10 seconds to an existing clip with WAN 2.5 Video Extend. 480p–1080p, billed only on the added duration from 10 credits/s, optional audio upload.
Video Extend adds 3, 5 or 10 seconds to a clip you already have. Built on Alibaba's WAN 2.5, it reads a prompt describing what happens next and continues the shot with matching motion, lighting and framing. Output is 480p, 720p or 1080p, and billing is based only on the added duration — from 10 credits per second — not on the length of the input.
- Input: Video + prompt
- Extension: 3 / 5 / 10s
- Resolutions: 480p–1080p
- Price: 10–30 credits/s
How to use Video Extend
Test the continuation cheaply at 480p, then re-run the winning prompt at the resolution the final cut needs.
- Prepare the source clip Trim to the point where the continuation should begin, remove fades and keep the original pacing. The final frame is the starting condition.
- Write only the next action Include the subject's movement and the camera behaviour, and match the energy of the source. Re-describing the input wastes the prompt.
- Run the cheapest useful test Three seconds at 480p costs 30 credits. Use it to check whether the direction and pacing work before paying for 1080p.
- Decide on audio before choosing length Upload a track only if it fits the output duration. A 30-second music bed on a 3-second extension will be cut after three seconds.
- Re-run at the final resolution Keep the seed and prompt that worked, then move to 720p or 1080p. Chain further runs only after reviewing the join.
Settings reference
Video Extend is WAN 2.5 continuation with four choices: extension length, resolution, an optional audio track and a seed. Billing covers the added seconds only.
| Setting | Value | Notes |
|---|---|---|
| Source video | Required, up to 100MB | The model continues from the clip's final frame. Trim to a clean ending before uploading. |
| Prompt | Required | Describe the next action and the camera behaviour. Do not re-describe what already happened. |
| Extension length | 3s / 5s / 10s | The default is 5 seconds. A 3-second run is the cheapest way to test a direction. |
| Resolution | 480p / 720p / 1080p | 720p is the default. Rates are 10, 20 and 30 credits per added second. |
| Audio upload | Optional WAV or MP3, 3–30s, up to 15MB | Longer audio is trimmed to the output length; shorter audio leaves the rest silent. |
| Seed | Optional | Fix a value to repeat a continuation after changing the prompt. |
| Price | 10 / 20 / 30 credits per added second | A 5-second 720p extension is 100 credits; the input video length is not billed. |
| Billing basis | Output duration only | A long source and a short source cost the same for the same extension and resolution. |
How extension works
The model treats the end of your clip as the starting condition and generates forward from there. The prompt describes the continuation: a character turns and walks away, the camera pans right, smoke drifts across the frame. Because the new material is generated rather than looped, the extension can introduce new action while keeping the original shot's look.
This is useful when a generation stops too early, when a camera move needs a beat more, or when a social cut needs to reach a minimum length. It is not a way to create a sequel: the extension continues the same shot, with the same subject and scene, rather than cutting to a new setup.
- Continues from the clip's final frame
- Prompt describes the next action
- New motion, same scene and grade
- Output is a longer continuous clip
- Not a scene-to-scene transition tool
Pricing on added seconds
Billing uses the selected output duration and resolution, not the input video's length. 480p costs 10 credits per second, 720p costs 20 and 1080p costs 30. A 3-second extension is therefore 30, 60 or 90 credits depending on resolution; a 10-second extension is 100, 200 or 300.
The smallest job is a 3-second extension at 480p for 30 credits, which makes iteration cheap. If you are unsure how the continuation should go, test at 480p or 720p for 3 seconds, then re-run at the final resolution with the prompt that worked.
- 480p: 10 credits per added second
- 720p: 20 credits per added second
- 1080p: 30 credits per added second
- Input length is not billed
- 3s at 480p is the cheapest test
| Extension | 480p | 720p | 1080p |
|---|---|---|---|
| 3s | 30 | 60 | 90 |
| 5s | 50 | 100 | 150 |
| 10s | 100 | 200 | 300 |
Prompting the continuation
Describe what happens next, not what already happened. “She turns and walks toward the door, camera holds” gives the model a clear action and camera instruction. Re-describing the input wastes the prompt and can push the model to repeat or contradict what is already on screen.
Match the original shot's energy. A slow, locked-off clip continued with a fast camera move will feel like two shots stitched together. If the clip ends mid-action, complete that action first; if it ends at rest, start a new beat. Keep the prompt short enough that the model can honour all of it.
- Describe the next action only
- Include camera behaviour
- Match the original pacing
- Complete mid-action cuts first
- Short prompts are easier to honour
Audio options
You can upload a WAV or MP3 track up to 15MB and 3–30 seconds to accompany the extension. If the audio is longer than the selected output length, only the first segment is used. If it is shorter, the remaining video plays silently. Without an upload, the model may generate audio or leave the result silent.
Choose the output duration to fit the audio when the track matters. A 10-second music bed with a 3-second extension will be cut after three seconds; a 3-second stinger with a 10-second extension leaves seven seconds of silence. For voice, generate the extension first and fit the audio to the final length rather than the other way around.
- WAV / MP3, ≤15MB, 3–30s
- Longer audio is trimmed to output length
- Shorter audio leaves silence
- No upload may generate audio or be silent
- Match output duration to the track
Limits and alternatives
The input video must be no more than 100MB, and a prompt is required. Output resolutions are 480p, 720p and 1080p; extension lengths are 3, 5 and 10 seconds. A seed can be fixed for repeatable results. Extensions can be chained — extend, then extend the result — with each run billed on its own added seconds.
HappyHorse 1.0 Video Extend offers more length choices (3, 5, 8, 10 and 15 seconds) at 720p or 1080p, priced at 50 and 100 credits per added second. WAN 2.2 Video Edit changes existing footage rather than lengthening it, and Video Upscaler raises resolution without changing duration.
- Input ≤100MB
- Prompt required
- 3 / 5 / 10 second extensions
- Seed supported
- Chain runs, each billed separately
Prompt examples for Video Extend
- Character exits: The woman turns and walks toward the door on the right, camera holds steady, soft afternoon light, natural pace. Complete the action the clip was already starting.
- Camera pan: Camera slowly pans right across the workshop, revealing tools on the bench, warm window light, smooth steady movement. Match the speed of the original move.
- Weather builds: Rain intensifies, droplets run down the window, the street reflections ripple, camera remains locked off. A good continuation for an atmospheric shot.
- Product rotate: The product continues its slow rotation, light sweeps across the surface, clean studio background, no cuts. Keep the motion continuous with the input.
- Crowd reaction: The crowd raises their hands and cheers, confetti falls, camera pushes in slightly, evening stadium light. Start the new beat at the clip's end.
- Quiet ending: The character lowers their head and stands still, wind moves through the grass, camera slowly pulls back, dusk light. Useful for landing a scene instead of cutting.
Pro tips for Video Extend
WAN 2.5's advantage is price and a 480p tier. Use the cheap tier for decisions and the expensive tier for delivery.
- Input length is never billed Only the selected output duration counts. There is no cost penalty for extending a long clip rather than a short one.
- Size the audio to the output Audio longer than the extension is trimmed; audio shorter than the extension leaves silence. Match the track length to the added seconds.
- The 30-credit test is the point A 3-second 480p run is the cheapest iteration in the video suite. Find the continuation there before spending 300 on a 10-second 1080p run.
- Match the original shot's energy A locked-off clip continued with a fast camera move reads as two shots. Complete a mid-action cut first, then start a new beat.
- Chained runs are billed separately Each extension is its own job at its own rate. Review every seam because drift accumulates across runs.
Frequently asked questions
How much does Video Extend cost?
10 credits per added second at 480p, 20 at 720p and 30 at 1080p. A 5-second 720p extension is 100 credits.
Is billing based on the input video length?
No. Only the selected output duration is billed. A long input clip and a short one cost the same for the same extension length and resolution.
What extension lengths are available?
3, 5 and 10 seconds.
What resolutions are supported?
480p, 720p and 1080p.
Is a prompt required?
Yes. The prompt describes what should happen in the added seconds.
Can I add my own audio?
Yes, a WAV or MP3 up to 15MB and 3–30 seconds. Longer audio is trimmed to the output length; shorter audio leaves the remaining video silent.
What is the input video limit?
Up to 100MB.
How does it differ from HappyHorse Video Extend?
HappyHorse offers 3, 5, 8, 10 and 15 second extensions at 720p or 1080p, priced at 50 and 100 credits per added second. WAN 2.5 is cheaper and adds a 480p tier, but caps the extension at 10 seconds.
Compare with other video tools
- HappyHorse 1.0 Video Extend
- WAN 2.2 Video Edit
- Kling Video Edit
- Video Upscaler
- Video Upscaler Pro
- Browse all AI video models
Specifications
- Provider: Wavespeed
- Model type: Video editing
- Credit pricing: 480p: 10 credits · 720p: 20 credits · 1080p: 30 credits
- Capabilities: audio-sync, multi-shot
About Conut.ai
Conut.ai is an AI image, video and audio generation platform with 50+ creative tools. It supports text-to-image, image-to-video, face swap, motion control, relighting, inpainting and video effects through models including Nano Banana Pro, Seedream, GPT Image, Sora 2, Kling, WAN and Seedance. Start free — no credit card required.