Wan 3.0 · One Video Model, Three Ways In
Alibaba's Tongyi Wanxiang Wan 3.0, on OVOV. Start from a prompt, from a first and last frame, or from a prompt backed by your own reference images, videos and audio — at 480P, 720P or 1080P, 5 to 30 seconds, with a native soundtrack.
AI video examples on OVOV
Real generations from our video models — Veo 3.1, Kling and Seedance. Hover any clip to play, click the speaker icon to hear the audio.
Hover to play · Click speaker to unmute
AI video examples on OVOV
Real generations from our video models — Veo 3.1, Kling and Seedance. Hover any clip to play, click the speaker icon to hear the audio.
Alien Fantasy World
Luxury Perfume Ad
Product Launch Teaser
Maldives Travel Commercial
Eastern Wuxia Duel
Hollywood Racing Movie
Spaceship to the Stars
Golden Retriever & Butterflies
Sushi Chef at Work
Red Dress on the Cliffs
Midnight Duel
Handheld Selfie Vlog
Mouth-watering Food Close-up
Urban Street Dance
Luxury Lipstick Ad
Why Wan 3.0 is worth a look
One model that takes text, keyframes or your own footage, renders up to 1080P, runs to half a minute, and writes its own soundtrack while it works.
Multimodal References
In text-to-video you can attach up to 10 reference images, up to 5 reference videos (15 seconds total) and up to 5 reference audio clips (15 seconds total). Wan 3.0 pulls character, style, motion and tone from your own material instead of guessing.
First & Last Frame Control
Give it a starting image, and optionally an ending image, and Wan 3.0 renders the motion between them. The straightest route to product reveals, transformations and before-and-after arcs.
480P, 720P and 1080P
Three resolution tiers, including a true 1080P output. Rough the idea out at 480P for a few credits, then render the version you actually ship at 1080P.
5 to 30 Seconds
Six fixed lengths: 5, 10, 15, 20, 25 or 30 seconds. A five-second test costs pocket change; a full 30-second spot fits inside one generation.
Native Audio, On by Default
Wan 3.0 writes ambience, effects and motion-matched sound alongside the picture. The switch is yours to flip, and turning audio off does not change the price.
Five Aspect Ratios, Priced by the Second
16:9, 9:16, 1:1, 4:3 and 3:4. Credits are charged per second — 3 at 480P, 6 at 720P, 12 at 1080P — and a failed generation is refunded in full.
From prompt, keyframes or footage to a finished clip
Pick the input that matches what you already have. Wan 3.0 handles picture and sound in the same pass, so there is nothing to assemble afterwards.
Choose your way in
Write a prompt on its own, upload a first frame (and optionally a last frame), or attach reference images, videos and audio alongside your prompt. Then set length, aspect ratio and resolution.
Wan 3.0 renders picture and sound
The model plans motion, lighting and pacing across the whole duration, then renders the video with its matching audio track. Most jobs land in one to five minutes; longer clips take longer.
Download and use it
Take the finished video with audio baked in — no watermark, commercial rights included. If a generation fails, your credits come straight back.
What Wan 3.0 is good for
Three input modes and a 1080P tier cover a lot of ground: finished ads, vertical social cuts, continuity work built on your own footage, and long atmospheric films.
1080P Product & Brand Spots
A 30-second commercial at full HD, rendered in one pass. Setup, demonstration, benefit and sign-off fit inside a single generation, sound included.
- 1080P output for pre-roll, landing pages and client delivery
- 16:9 for widescreen, 4:3 for classic broadcast framing
- Product sound and ambience generated with the picture
Vertical Social Shorts
TikTok, Reels and Shorts in 9:16, from a 5-second hook to a 30-second story. Audio arrives attached, so nothing needs a soundtrack pass afterwards.
- 9:16 vertical, or 1:1 for square feed posts
- Draft at 480P for 3 credits a second, publish at 720P or 1080P
- Native audio instead of a licensed music track
Consistency Across a Series
Attach the same reference images, video and audio to every generation and the character, product and tone carry from clip to clip instead of drifting.
- Up to 10 reference images to hold character and style
- Up to 5 reference videos for motion and camera behaviour
- Up to 5 reference audio clips to keep the sound in family
Travel & Atmosphere Films
Aerial sweeps, coastal drives and street-level wandering that need time to breathe — plus the wind, water and city sound that sells the place.
- Long drifting camera moves that need the full 30 seconds
- Environmental audio rendered alongside the visuals
- 1080P for hero videos that sit at the top of a page
Prompt writing guide for Wan 3.0
Wan 3.0 accepts prompts up to 20,000 characters, which is far more room than most clips need. These habits spend that room on the things the model actually acts on.
Name your materials, then describe the finished shot
Pick the input mode before you write
The three modes take different prompts. Text-to-video has to carry the whole scene in words. First-and-last-frame already has the composition, so the prompt is about movement. Reference mode is about what to borrow from the material you attached.
- •Text only: describe setting, subject, camera and sound from scratch
- •First and last frame: describe the motion between them, not the frames themselves
- •References (attached in text-to-video): say what each asset contributes — face, outfit, camera move, tone
Example
"First frame: a sealed cardboard box on a table. Last frame: the box open with a camera inside. Prompt: "hands lift the flaps in one smooth motion, soft window light"."
Address your references by number
In reference mode you can point at the material directly — image 1, video 1, audio 1 — so the model knows which asset governs which part of the shot instead of blending everything together.
- •Up to 10 reference images, 5 reference videos and 5 reference audio clips
- •Reference video totals 15 seconds; reference audio totals 15 seconds
- •Give each asset one job: identity from one image, camera move from one video
Example
"Keep the woman from image 1 and the jacket from image 2, follow the slow dolly-in of video 1, and match the room tone of audio 1."
Plan the arc across the length you chose
Thirty seconds is a lot of screen time. Write the progression in the order it should happen, or the model will hold a single idea for half a minute.
- •Open with the setting, bring in the subject, then the movement
- •Name what changes over time: light, weather, distance, speed
- •Shorter lengths (5-10s) suit one continuous action
Example
"The room starts dim and empty, morning light climbs the far wall, dust drifts through the beam, and the camera slowly settles on a cup of coffee going cold."
Write the sound, not just the picture
Audio is generated by default, so anything you leave unsaid gets invented for you. One or two lines of sound direction usually keeps it in character.
- •Name the ambience: rain on glass, distant traffic, room tone
- •Name the sounds that match the motion: footsteps, engine, wind
- •Asking for quiet is valid direction — say so if you want it sparse
- •Attach reference audio when you already have a tone in mind
Example
"Sound: steady rain on a tin roof, an occasional low roll of thunder in the distance, no music."
Trade praise words for visual facts
Stunning, epic and cinematic tell the model nothing. Materials, colours, light direction and strong verbs do.
- •Skip: stunning, epic, breathtaking, gorgeous, masterpiece
- •Use verbs with force: slams, drifts, ignites, whips
- •Name the light source and the time of day
Example
"Avoid: "a stunning epic mountain shot". Prefer: "granite ridges under low side light, snow spilling off the crest in a thin plume"."
Choose length, ratio and resolution upfront
Wan 3.0 offers 5, 10, 15, 20, 25 and 30 seconds, five aspect ratios and three resolutions. Deciding before you generate saves credits and keeps the composition right.
- •16:9 and 4:3 for landscape, 9:16 and 3:4 for vertical, 1:1 for feeds
- •Draft at 480P (3 credits per second), finish at 720P (6) or 1080P (12)
- •Write for the frame you picked — vertical needs the subject close and centred
- •Length is fixed to 5, 10, 15, 20, 25 or 30 seconds — round your storyboard to one of them
Example
"A 15-second 9:16 vertical clip: a barista pulling an espresso shot, steam rising into warm café light, shallow depth of field."