Technologia WAN 2.7 LoRA umożliwia tworzenie filmów z zachowaniem charakteru, stylu i motion control zgodnej z technologią LoRA, co pozwala na uzyskanie spójniejszych i bardziej kinowych rezultatów.
curl-X POST 'https://pollo.ai/api/platform/v1/generation/alibaba/wan-v2-7-lora/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", "image": "https://example.com/first-frame.jpg", "duration": 5, "resolution": "720p" }}'
Opcjonalnie dołącz webhookUrl, aby otrzymać wywołanie zwrotne w przypadku powodzenia lub niepowodzenia zadania.
Opcjonalne pole żądania
{"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","image":"https://example.com/first-frame.jpg","duration":5,"resolution":"720p"},"webhookUrl":"https://example.com/webhooks/pollo"}
Schemat
Pola wejściowe dla wybranego punktu końcowego i trybu.
Pole
Typ
Wymagany
Opis
prompt
string
Tak
Text prompt describing the content to generate.
negativePrompt
string
NIE
Text describing elements to avoid in the generated output.
image
string
Tak
First-frame / reference image URL (HTTP(S).
audio
string
NIE
Reference audio URL (HTTP(S)).
duration
integer
NIE
The duration of the generated video, in seconds.
resolution
string
NIE
Output resolution of the generated media (e.g. 720p, 1080p, 2K, 4K). Values are case-insensitive.
seed
integer
NIE
Random seed for reproducible generation.
Błędy
Typowe publiczne kody odpowiedzi API.
400
Nieprawidłowe dane wejściowe lub ograniczenie modelu
Wan 2.7 LoRA supports LoRA-guided video generation for character, style, and motion consistency. It helps apply a chosen visual direction across related clips, making series-based outputs easier to keep aligned.
With Wan 2.7 LoRA on Pollo API, you generate from an image and prompt together, adding optional audio-driven motion, choosing 720P or 1080P, and lengths from 2 to 15 seconds per request.
Key Features of Wan 2.7 LoRA API
LoRA-Driven Consistency
The model applies LoRA conditioning to hold a chosen character, subject, or style across separate generations, so a recurring look survives from one clip to the next.
Image-Anchored Motion
Every generation starts from a source image, letting Wan 2.7 LoRA animate an existing frame while respecting its composition, colors, and subject rather than reinventing the scene.
Audio-Driven Movement
Supply an audio track and the model can time motion to that sound, aligning gestures and pacing with a voice, beat, or ambient cue for synced results.
Flexible Clip Length
Durations run from two-second beats to fifteen-second sequences, giving room for both quick loops and longer narrative shots without stitching separate outputs together.
Prompt and Negative Control
Positive prompts steer the action while negative prompts prune unwanted elements, giving tighter control over what appears and what stays out of frame.
Use Cases of Wan 2.7 LoRA API
Recurring Character Series: Reuse a trained LoRA to keep the same mascot or avatar on-model across an entire episode set or ad campaign.
Branded Style Locking: Pin a signature color grade and art direction so every generated clip reads as one brand, not a mix of looks.
Lip-Synced Talking Clips: Pair a portrait image with a voice track to drive mouth and head motion timed to the supplied audio.
Music-Reactive Visuals: Feed a beat-driven track and animate a still into motion that pulses and moves in time with the sound.
Product Loop Animation: Turn a packshot image into a short repeating motion clip for storefronts, keeping the item's exact shape and finish.
Consistent Explainer Scenes: Generate a sequence of shots that share one narrator or setting, useful for tutorial and onboarding tools.
Stylized Fan Content: Apply a distinct illustration LoRA to animate character art for communities built around a specific visual identity.
How to Use Wan 2.7 LoRA API
Get an API Key: Create a Pollo API account and generate your API key from the developer dashboard.
Choose the Model: Send your request to the Wan 2.7 LoRA endpoint at /generation/wanx/wan-v2-7-lora.
Add Your Inputs: Provide a source image URL and a prompt, then optionally set audioUrl, negativePrompt, resolution (720P or 1080P), and length (2 to 15).
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 LoRA API
Describe the subject in your source image, the motion you want added, and the style you expect the LoRA to hold, so guidance and animation pull in the same direction.
A Simple Prompt Formula
Subject from source image + specific motion or action + emotional or timing cue + style to preserve + what to exclude
What You Should Notice
Match Prompt To Image: Reference what the source frame actually shows so the added motion feels native to that subject.
Name The Motion: State the exact movement, such as turning, nodding, or drifting, to give the animation a clear target.
Lean On Negatives: Use negative prompts to remove artifacts, extra limbs, or unwanted background clutter.
Sync To Audio: When using an audio track, describe how motion should follow the sound for tighter timing.
Right-Size The Length: Pick shorter durations for loops and longer ones for sequences that need a small arc.
Example Prompts
Consistent Avatar Motion
"Animate the illustrated fox character from the source image waving one paw and blinking twice, keeping its exact fur pattern and flat pastel style, gentle idle sway, soft studio background, no extra characters in frame."
Audio-Synced Portrait
"The person in the source photo speaks warmly to camera, lips and head timed to the provided voice track, calm natural expression, steady framing, preserve the original lighting and skin tone, no background movement."
Product Loop
"Slowly rotate the ceramic mug from the source image on a matte surface, steam rising evenly, keep the glaze color and printed logo intact, soft even light, short seamless loop, no reflections or clutter."
Music-Reactive Scene
"Animate the neon geometric artwork from the source image pulsing and shifting in time with the supplied beat, colors brightening on each hit, smooth looping motion, dark backdrop, preserve the original shapes and palette."
Wan 2.7 LoRA vs Kling 2.6 vs Vidu Q1
Capability
Wan 2.7 LoRA
Kling 2.6
Vidu Q1
LoRA style/character locking
✅ Native LoRA conditioning
❌ No LoRA control
❌ No LoRA control
Image-anchored generation
✅ Required source image
✅ Image-to-video
✅ Reference-to-video
Audio-driven motion
✅ Optional audio track
❌
✅
Resolution options
720P and 1080P
Up to 1080P
Up to 1080P
Clip length
2 to 15 seconds
Short directed clips
Short clips
Recommended For
Consistent recurring characters and styles
Fluid general motion
Reference-guided scenes
Why Choose Wan 2.7 LoRA API?
Wan 2.7 LoRA fits products that need the same character or style to reappear reliably, pairing LoRA conditioning with image-anchored, optionally audio-driven video generation.
Through Pollo API, you reach Wan 2.7 LoRA 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 LoRA endpoint on Pollo API to ship consistent, style-locked video features, then swap to other models without rewriting your integration.
Wan 2.7 LoRA API FAQs
What is Wan 2.7 LoRA?
Wan 2.7 LoRA is a wanx video model that adds LoRA-based character, style, and motion guidance to image-anchored generation, keeping a chosen look consistent across clips.
Does Wan 2.7 LoRA API need an image input?
Yes. Each request takes a source image plus a text prompt, so the model animates and guides an existing frame rather than generating from text alone.
Can it use audio to drive motion?
Yes. You can supply an audio track so movement and timing follow the sound, which suits talking portraits and music-reactive clips.
What resolutions and lengths are supported?
Wan 2.7 LoRA outputs 720P or 1080P video, with clip lengths selectable from 2 up to 15 seconds per request.
How does LoRA guidance help?
LoRA conditioning locks a specific character, subject, or style, so repeated generations stay on-model instead of drifting between different looks.
Why run Wan 2.7 LoRA through Pollo API?
Pollo API gives one integration for Wan 2.7 LoRA and 300+ other models, with documentation, task tracking, and lower-cost generation than comparable providers.
