WAN 2.6 Image-to-Video
Animate a photo with WAN 2.6 image-to-video: 720p/1080p, 5–15s clips, native sound, 30–50 credits/s. Longer cinematic shots with Alibaba's WAN 2.6 model.
WAN 2.6 Image-to-Video extends Alibaba's image animation line to 15-second clips and 1080p, with sound generated alongside the picture. It keeps the source image's composition and identity while adding cinematic camera motion, natural movement and a longer story beat than the 5- and 8-second WAN 2.2 models. On Conut.ai it costs 30 credits per second at 720p and 50 credits per second at 1080p; generation is slower than the smaller models, with an estimated five minutes per job.
- Resolution: 720p / 1080p
- Duration: 5s / 10s / 15s
- Price: 30–50 credits/s
- Audio: Native sound
How to use WAN 2.6 Image-to-Video
The longer runtime is the main feature. Build the prompt as a sequence of beats rather than one continuous action.
- Upload a source image with depth Fifteen seconds of motion needs somewhere to go. A frame with foreground, subject and background gives the camera room to move.
- Write the prompt as beats For a 15-second clip, describe three moments: what happens first, what changes in the middle, and how it lands. One long sentence tends to produce one long drift.
- Match duration to the story Five seconds for a gesture, ten for a short sequence, fifteen for a mini-scene. Paying for fifteen seconds to show one movement wastes credits.
- Decide on 720p or 1080p 720p is half the per-second price and the default. Move to 1080p for hero shots and anything with fine detail or text.
- Describe the sound in the prompt Since audio is generated with the picture, name the ambience and key effects you expect. The model has no separate audio field to guide it.
Settings reference
WAN 2.6 Image-to-Video extends the line to 15-second clips and 1080p, with sound generated alongside the picture. The control set stays minimal.
| Setting | Value | Notes |
|---|---|---|
| Input image | 1 image, required | The source fixes identity and composition for the whole clip. |
| Prompt | Required | Describe motion, story beats, camera moves and style. |
| Duration | 5s / 10s / 15s | Default is 5 seconds. Fifteen seconds needs material for three beats. |
| Resolution | 720p / 1080p | 30 credits per second at 720p, 50 at 1080p. 720p is the default. |
| Native sound | Generated with the picture | Ambience and effects come from the prompt; there is no separate audio upload on this endpoint. |
| Seed | Optional | Reuse a seed to reproduce a take while you adjust the prompt. |
| Negative prompt | Not exposed | This endpoint has no negative prompt or prompt-expansion toggle. |
What changes in WAN 2.6
The headline change is duration. WAN 2.6 can run for 15 seconds in a single job, which is long enough for a three-beat mini-scene instead of a single gesture. It also starts at 720p and goes to 1080p, so it is aimed at finished footage rather than low-resolution drafts. Sound is generated with the clip, which makes it a better fit for trailers, product films and social videos that need ambience without a separate audio pass.
The trade-off is cost and waiting time. There is no 480p tier, and the 1080p rate is 50 credits per second, so a 15-second 1080p clip costs 750 credits. Renders are estimated at around five minutes, which is normal for a model producing three times the frames of a 5-second WAN 2.2 clip.
- Up to 15 seconds per generation, with 5- and 10-second options
- 720p and 1080p output; no 480p draft tier
- Sound is generated with the video
- Estimated five minutes per render
- Preserves the source image's composition and identity
WAN 2.6 vs WAN 2.5 vs WAN 2.2
WAN 2.6 is the longest and highest-resolution WAN image-to-video model on Conut.ai. WAN 2.5 is cheaper and accepts an uploaded audio track; WAN 2.2 is the low-cost silent option. The right choice depends on whether the constraint is duration, sound, control or budget.
- Use WAN 2.6 when the shot needs 10–15 seconds or 1080p.
- Use WAN 2.5 when you need to upload a specific voice track.
- Use WAN 2.2 Spicy for cheap, expressive 5-second drafts.
- Use standard WAN 2.2 when you need a negative prompt or last frame.
| Model | Resolution | Duration | Credits | Audio | Best for |
|---|---|---|---|---|---|
| WAN 2.2 I2V | 480p / 720p | 5s, 8s | 30–60 /s | No | Controlled silent motion |
| WAN 2.2 Spicy | 480p / 720p | 5s, 8s | 20–40 /s | No | Expressive silent motion |
| WAN 2.5 I2V | 480p / 720p / 1080p | 5s, 10s | 20–40 /s | Yes, audio upload | Talking shots, custom voice |
| WAN 2.6 I2V | 720p / 1080p | 5s, 10s, 15s | 30–50 /s | Native sound | Longer cinematic clips |
Pacing a 15-second shot
A 15-second clip is not a 5-second clip stretched three times. It needs a beginning, a change and an ending. The simplest structure is three beats of roughly five seconds each: establish the subject and setting, introduce the action or camera move, then land on a final composition. Say what happens in each beat instead of listing every object in the scene.
Keep one dominant action across the clip and let the camera carry the continuity. If the subject has to do two unrelated things, the model often rushes the second one or drops it. For product films, a reveal followed by an orbit is enough. For character shots, a turn followed by a step toward camera reads better than a full choreography.
- Write three beats: establish, change, resolve.
- Keep one dominant subject action; let the camera move through the beats.
- Avoid cutting between unrelated locations inside one generation.
- Match the source image to the first beat, not the middle of the action.
- For 5-second tests, reuse the first beat of the 15-second prompt.
Sound, resolution and cost
WAN 2.6 generates sound with the clip, so prompts can describe ambience, weather, footsteps or machinery. This endpoint does not expose an audio upload field; if you need a specific voice or music track, generate the video here and add the track in an editor, or use WAN 2.5 image-to-video, which accepts WAV and MP3 files.
Resolution and duration multiply quickly. A 5-second 720p clip is 150 credits, while a 15-second 1080p clip is 750. Draft the motion and camera at 720p for 5 seconds, then re-run the winning prompt at the final resolution and duration.
- 720p: 30 credits/s; 1080p: 50 credits/s
- 5s 720p = 150 credits; 10s 720p = 300; 15s 720p = 450
- 5s 1080p = 250 credits; 10s 1080p = 500; 15s 1080p = 750
- Describe ambience in the prompt: wind, rain, crowd, machinery, room tone
- For a specific voice track, pair with WAN 2.5 or add audio in post
Limitations and failure cases
WAN 2.6 is not the model for quick iteration. Five-minute renders and a 30-credit-per-second floor make it a finishing tool, not a sketchpad. It also has no 480p option, so cheap low-resolution tests are not available. Faces, hands, text and fine patterns remain the weak points, and the longer the clip, the more chances the model has to drift.
- Draft with WAN 2.2 or WAN 2.5, then finish with WAN 2.6.
- Avoid text and logos that must stay readable for 15 seconds.
- Keep faces and hands large enough to hold detail, or keep them still.
- Long clips can lose lighting consistency; describe one light setup.
- If the shot is under 8 seconds and silent, a cheaper model is usually enough.
Prompt examples for WAN 2.6 Image-to-Video
- Post-apocalyptic walk: A lone soldier in a weathered coat walks slowly toward the camera through a ruined street, warm orange sunset light, dust in the air. Beat 1: wide shot, establish the empty street. Beat 2: slow dolly-in as the soldier passes a burnt-out car. Beat 3: close-up on the helmet as the sun drops behind the buildings. Cinematic, shallow depth of field, ambient wind and distant debris. Three beats, one walking action, one continuous camera move.
- Product reveal: A brushed metal watch on black stone, single hard spotlight. Beat 1: macro on the dial. Beat 2: camera orbits 45 degrees as the second hand moves. Beat 3: pull back to reveal the full watch and its reflection. Luxury commercial lighting, crisp details, subtle room tone. A rigid product hides motion artifacts and makes the three beats easy to read.
- City timelapse: A rooftop view of a city at dusk. Beat 1: clouds drift over the skyline. Beat 2: office lights switch on across the buildings as the camera slowly pushes in. Beat 3: the first stars appear above the towers. Documentary timelapse feel, deep blue and amber palette, distant traffic ambience. Landscapes and cityscapes tolerate long camera moves without a subject to warp.
- Fantasy creature: A white stag with glowing antlers stands in a misty forest clearing. Beat 1: the stag lifts its head. Beat 2: fireflies rise around it as the camera orbits slowly. Beat 3: the stag turns and walks into the fog. Cinematic fantasy realism, volumetric moonlight, soft forest ambience. One creature action per beat keeps the longer clip coherent.
- Fashion walk: A model in a long red coat walks through a rain-soaked street at night. Beat 1: wide shot, neon reflections on the pavement. Beat 2: tracking shot alongside the model as the coat moves. Beat 3: the model stops under a streetlamp and looks toward the camera. Cinematic fashion film, high contrast, rain and distant traffic. A tracking shot carries continuity across the three beats.
- Documentary portrait: An elderly fisherman mends a net on a wooden dock at dawn. Beat 1: close-up on weathered hands. Beat 2: camera slowly pulls back to reveal the harbor behind him. Beat 3: he looks up at the sunrise. Natural documentary style, soft morning light, gulls and water ambience. Small human actions are easier to keep stable than fast gestures.
Pro tips for WAN 2.6 Image-to-Video
Longer clips reward planning. Most weak 2.6 results are 15-second runs with a 5-second idea.
- Fifteen seconds needs three beats Write an opening state, a change and an ending state. If you cannot name all three, use a shorter duration and keep the clip tight.
- Check the cost before a long run A 15-second 1080p clip costs 750 credits. Test the motion at 5 seconds and 720p first, then extend only the take that works.
- Keep identity with a clear source The model holds the source image's composition for longer, so a sharp, undistorted photo matters more than prompt detail.
- Use 2.5 when dialogue is the point WAN 2.5 Image-to-Video supports audio upload and is built around one-pass speech sync. WAN 2.6 is better for cinematic sound and longer silent beats.
- Direct the camera explicitly Slow push-in, lateral tracking, static frame: naming the move keeps a long clip from drifting into aimless motion.
Frequently asked questions
How long can a WAN 2.6 image-to-video clip be?
Up to 15 seconds per generation. The endpoint supports 5-, 10- and 15-second durations.
Does WAN 2.6 generate sound?
Yes. Sound is generated with the clip. This endpoint does not expose an audio upload field; use WAN 2.5 image-to-video if you need to supply a specific voice track.
How much does WAN 2.6 image-to-video cost?
720p costs 30 credits per second and 1080p costs 50 credits per second. A 15-second 720p clip is 450 credits and a 15-second 1080p clip is 750 credits.
Can WAN 2.6 output 480p?
No. The endpoint supports 720p and 1080p only. Use WAN 2.5 or WAN 2.2 for cheaper low-resolution drafts.
Why does WAN 2.6 take longer than other models?
It generates more frames at higher resolutions. The page estimates around five minutes per job, especially for 15-second 1080p clips.
What is the difference between WAN 2.5 and WAN 2.6?
WAN 2.5 is cheaper, supports 480p drafts and accepts uploaded audio, but caps clips at 10 seconds. WAN 2.6 adds 15-second clips and 1080p, with sound generated by the model.
Does WAN 2.6 support a negative prompt or last frame?
This endpoint exposes a seed and native sound, but not a negative prompt or last-frame image. For those controls, use standard WAN 2.2 image-to-video or WAN 2.5.
What images work best with WAN 2.6?
A sharp, well-lit image with one clear subject and a simple background. The model preserves the source composition for the full clip, so a cluttered or soft input limits the result at every duration.
Compare with other image-to-video models
- WAN 2.5 Image-to-Video
- WAN 2.2 Image-to-Video
- WAN 2.2 Spicy
- Seedance 2.5 Image-to-Video
- Kling O3 Pro Image-to-Video
- Browse all AI video models
Specifications
- Provider: Wavespeed
- Model type: Image to video
- Credit pricing: 720p: 30 credits · 1080p: 50 credits
- Capabilities: new, cinematic, starter, 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.
Explore more AI tools
- Seedance 2.5 Image-to-Video
- Seedance 2.5 Turbo Image-to-Video
- Seedance 2.0 Mini Text-to-Video
- Seedance 2.0 Mini Image-to-Video
- MiniMax H3 Text-to-Video
- MiniMax H3 Image-to-Video
- MiniMax H3 Reference-to-Video
- Wan 3.0 Text-to-Video
- Wan 3.0 Image-to-Video
- Photo Restore
- Cinema Studio
- Viral AI Influencer