Wan 2.7 создает многокадровые кинематографические видеоролики со стабильными персонажами, естественным движением и последовательным, продолжительным повествованием.
curl-X POST 'https://pollo.ai/api/platform/v1/generation/alibaba/wan-v2-7/video'\-H'Content-Type: application/json'\-H'x-api-key: YOUR_API_KEY'\-d'{ "input": { "prompt": "A cinematic shot of a golden retriever running through a field of sunflowers at sunset, warm rim light, shallow depth of field", "negativePrompt": "blurry, low quality, distorted, watermark, text", "duration": 5, "resolution": "720p", "aspectRatio": "1:1" }}'
При желании укажите webhookUrl, чтобы получать обратный вызов в случае успешного или неудачного выполнения задачи.
Необязательное поле запроса
{"input":{"prompt":"A cinematic shot of a golden retriever running through a field of sunflowers at sunset, warm rim light, shallow depth of field","negativePrompt":"blurry, low quality, distorted, watermark, text","duration":5,"resolution":"720p","aspectRatio":"1:1"},"webhookUrl":"https://example.com/webhooks/pollo"}
Схема
Поля ввода для выбранной конечной точки и режима.
Поле
Тип
Необходимый
Описание
audio
string
Нет
Reference audio URL (HTTP(S)).
prompt
string
Да
Text prompt describing the content to generate.
negativePrompt
string
Нет
Text describing elements to avoid in the generated output.
duration
integer
Нет
The duration of the generated video, in seconds.
resolution
string
Нет
Output resolution of the generated media (e.g. 720p, 1080p, 2K, 4K). Values are case-insensitive.
aspectRatio
string
Нет
Output aspect ratio, as width:height (e.g. 16:9).
seed
integer
Нет
Random seed for reproducible generation.
Ошибки
Типичные коды ответов общедоступных API.
400
Недопустимые входные данные или ограничения модели.
Wan 2.7 keeps a character recognizable when the scene cuts. It carries faces, wardrobe, and setting across multiple shots, so a short sequence reads as one continuous story rather than disconnected clips.
Through Wan 2.7 on Pollo API, developers can generate from text or from a start frame, guide motion with audio, set clip length from 2 to 15 seconds, and output at 720P or 1080P.
Key Features of Wan 2.7 API
Multi-Shot Consistency
Characters, costumes, and locations stay stable as shots change, letting you build a beginning, middle, and end that feel connected instead of restarting the look each cut.
First and Last Frame Control
Provide a starting image and an ending image, and Wan 2.7 fills the motion between them, giving you a defined arc for transitions, morphs, and before-and-after reveals.
Audio-Driven Motion
Supply an audio track and the model times movement to it, syncing gestures, pacing, and expression so speech and rhythm shape the action rather than running separately.
Extended Length Range
Choose any length from 2 to 15 seconds per request, so you can produce a quick loop or a longer narrative beat without stitching separate generations together.
Natural Camera Work
Pans, pushes, and orbits move with steady intent, and subjects hold their motion believably through the shot, keeping longer takes coherent from first frame to last.
Multilingual Prompting
Wan 2.7 reads prompts across languages and holds visual meaning steady, so teams working in different languages can describe the same scene and get consistent results.
Use Cases of Wan 2.7 API
Serialized Short Stories: Generate a character who reappears across several shots, useful for episodic tools where the same protagonist must stay recognizable scene to scene.
Start-to-End Transitions: Set a first and last frame to animate a sketch resolving into a finished mural, or a seedling opening into full bloom.
Narrated Explainers: Feed an audio track so an on-screen guide moves and gestures in time with the voiceover for tutorial and education apps.
Localized Ad Variants: Prompt the same scene in different languages to produce market-specific versions without redrawing the visual concept each time.
Longer Storyboard Beats: Use the wider length range to preview a full narrative moment, giving creative teams more than a two-second glimpse to judge pacing.
Image-Led Continuations: Start from a still and extend it into motion while keeping the subject and framing intact for asset-first workflows.
Music-Timed Loops: Align short clips to a beat with audio guidance, fitting rhythm-driven content and looping social formats.
Aspect-Aware Delivery: Generate the same idea in vertical, square, or widescreen framing to match feed, story, and landscape placements.
How to Use Wan 2.7 API
Get an API Key: Create a Pollo API account and generate your API key from the developer dashboard.
Choose the Model: Send requests to the Wan 2.7 endpoint at /generation/wanx/wan-v2-7.
Add Your Inputs: Pass a prompt for text-to-video, or an image (plus optional imageTail) for image-to-video, then set length (2-15), resolution (720P or 1080P), and audioUrl for audio-driven motion.
Generate and Retrieve: Submit the task, poll the returned task status, and download the finished video once it succeeds.
Prompting Best Practices for Wan 2.7 API
Describe your character and setting once with enough detail to anchor them, then lay out the shot sequence in order so Wan 2.7 knows what to keep steady across cuts.
A Simple Prompt Formula
Recurring subject with fixed details + shot-by-shot actions + camera movement + setting and mood + optional audio or language cue
What You Should Notice
Anchor The Character: Give consistent, specific traits early so the same person stays recognizable when the scene changes.
Order Your Shots: Write actions in sequence to guide how the multi-shot story unfolds from start to finish.
Frame The Arc: When using first and last frames, make the two images clearly related so the motion between them makes sense.
Time With Audio: Reference the pacing or rhythm when supplying an audio track so movement lines up with sound.
Match Length To Story: Pick a longer duration for narrative beats and a short one for loops and quick previews.
Set The Frame Shape: Choose the aspect ratio that fits your placement before generating a text-to-video clip.
Example Prompts
Recurring Character Sequence
"A young lighthouse keeper checks the lamp at dusk, then climbs down the spiral stairs, then steps onto the rocky shore as waves rise. Same face and coat throughout, cool blue light, slow steady camera following each move."
First-to-Last Frame Morph
"Start on a blank ceramic bowl, end on the same bowl fully glazed in deep indigo patterns. Hands paint and shape it in between, warm studio light, close overhead framing, smooth continuous motion."
Audio-Timed Narration
"A market vendor explains her spice blends to camera, hands lifting jars and pointing to labels in time with her voice. Bright stall backdrop, gentle push-in, gestures paced to the supplied audio track."
Longer Narrative Beat
"A hiker reaches a ridge, catches her breath, unfolds a paper map, then looks up as clouds part over the valley. Twelve-second single take, wide mountain light, camera drifting slowly to reveal the view."
Wan 2.7 vs Kling 2.6 vs Sora 2
Capability
Wan 2.7
Kling 2.6
Sora 2
Multi-shot character consistency
✅ Holds subject across cuts
✅ Strong within a shot
✅ Strong scene coherence
First and last frame control
✅
✅
❌
Audio-driven motion
✅
❌
✅ Generated audio focus
Clip length range
2-15 seconds
Short directed clips
Short clips
Resolution options
720P and 1080P
Up to 1080P
Up to 1080P
Multilingual prompting
✅
✅
✅
Recommended For
Multi-shot narratives with stable characters
Expressive character motion
Coherent scenes with generated audio
Why Choose Wan 2.7 API?
Wan 2.7 fits products that need a story, not a single shot, holding characters and settings steady across cuts while giving you frame control, audio timing, and a wide length range.
Through Pollo API, you reach Wan 2.7 with one API key alongside 300+ leading video and image models, with clean docs, task status polling, and faster generation.
Start with the Wan 2.7 endpoint on Pollo API to ship multi-shot, character-consistent video features, then compare it against Kling, Sora, and other models without changing your integration.
Wan 2.7 API FAQs
What is Wan 2.7?
Wan 2.7 is a video generation model built for multi-shot storytelling, keeping characters and scenes consistent across cuts while supporting text-to-video, image-to-video, and audio-driven motion.
Does Wan 2.7 API support text and image inputs?
Yes. You can generate from a text prompt or from a starting image, and you can add an ending image to define the motion between two frames.
What clip lengths and resolutions are available?
Wan 2.7 supports lengths from 2 to 15 seconds and outputs at 720P or 1080P, selected per request to match your use case.
How does audio-driven generation work?
Supply an audio URL and Wan 2.7 times movement to the track, syncing gestures and pacing so on-screen action aligns with the sound.
Can Wan 2.7 keep a character consistent across shots?
Yes. That is its core strength: the same character, wardrobe, and setting stay recognizable across multiple shots, so short sequences read as one connected story.
Why run Wan 2.7 through Pollo API?
Pollo API gives one integration for Wan 2.7 and 300+ other models, with documentation, task tracking, webhooks, and lower-cost generation than comparable providers.
