Choose a category
3 models available
4 models available
3 models available
2 models available
4 models available
8 models available
Wan 2.7 Text to Video
1080p native, five aspect ratios, optional audio track.
Creative Vision
Describe the scene, the camera and the motion.
Video Parameters
Cinema Preview
Generation History0 records
Wan 2.7 Text to Video Generator
Wan 2.7 text to video is the controllable option on this site: native 1080p output, negative prompts, a fixed seed for repeatable results, and 2 to 15 seconds per run. Generate from a prompt alone, or supply an audio track and have the motion follow it.
Best for shots you need to iterate on: same seed, one changed word, and a predictable difference in the result
Why Choose Wan 2.7 Text to Video
Alibaba's Wan family renders natively at 1080p and exposes the controls that most video models hide: what to exclude, which seed to reuse, and whether the prompt gets rewritten before it runs.
Native 1080p
720p for drafts, 1080p when the clip ships. 1080p is a genuine render rather than an upscale, and it is priced accordingly at about 5.4 credits per second of finished video.
Negative Prompts
Up to 500 characters of what you do not want: extra fingers, text overlays, camera shake, lens flare. This is the fastest fix for a shot that is almost right.
Seed Control
Reuse a seed and the same prompt lands close to the same result. Change one word, keep the seed, and you get a comparable variant instead of an unrelated clip.
Audio-Guided Generation
Upload a track up to 50MB and the video is generated to follow it. Useful when motion has to hit a beat or match the rhythm of a voiceover you already have.
Prompt Optimizer
On by default: Wan expands a short prompt into a fuller description before rendering. Turn it off when your prompt is already precise and you want it obeyed literally.
2 to 15 Seconds, Five Ratios
A per-second slider from 2 to 15, plus 16:9, 9:16, 1:1, 4:3 and 3:4. Short 2-second runs make cheap tests before you commit to a long take.
Iterate Cheaply, Render Once
Test the framing with a 2-second 720p run for 8 credits. Lock the seed, refine the negative prompt, then render the final take at 1080p.
How to Use Wan 2.7 Text to Video
The workflow that gets the most out of this model is deliberately iterative, because seed control makes iteration actually worth doing.
Write the Prompt
Describe the scene, the camera and the motion. Wan responds well to camera language: slow push in, handheld follow, static wide, rack focus.
Test Short and Cheap
Start at 2 seconds and 720p, which costs 8 credits. You are checking composition and subject, not the full performance, so length is wasted here.
Fix What Is Wrong
Add the problem to the negative prompt rather than rewriting the whole prompt. Note the seed of the run that came closest and reuse it.
Render the Final Take
Same prompt, same seed, now at full length and 1080p. Generation takes a few minutes at this tier, and the finished clip lands in your history.
A Note on the Prompt Optimizer
It helps short prompts and gets in the way of long ones. If a carefully written 300-word prompt comes back reinterpreted, switch the optimizer off and run it again.
Wan 2.7 Text to Video
FAQ
Resolutions, seeds, negative prompts, audio input and credit costs for the Wan 2.7 text-to-video model.
Wan 2.7 is a video generation model from Alibaba's Wan series. This page runs its text-to-video endpoint: you supply a prompt and settings, and we handle the generation, the credit accounting and the result storage. FlowVeo3 is an independent platform and is not affiliated with Alibaba.
Between 8 and 85 credits. Pricing is per second: about 3.6 credits per second at 720p and 5.4 at 1080p, rounded up to a clean figure. That makes 2 seconds at 720p 8 credits and a full 15 seconds at 1080p 85 credits. The exact cost is shown before you submit.
Because 1080p costs half again as much per second, and defaulting to the most expensive tier would charge people who never opened the resolution menu. Switch to 1080p deliberately, for the take you intend to keep.
It fixes the random starting point of the generation. Two runs with the same prompt, settings and seed produce near-identical video, so a seed lets you change one thing at a time and see what that change did. Leave it empty for a fresh result each run.
List concrete artefacts, not moods. "blurry, watermark, text, extra limbs, camera shake" works; "bad quality" does not. You have 500 characters, and short focused lists generally beat long ones.
The audio field is an input rather than an output setting. Upload a track and the video is generated to follow it, which is how you get motion that matches a beat or a voiceover. Do not count on a returned clip carrying its own generated audio: if you need specific sound, supply it.
Expect a few minutes, longer for 15-second 1080p runs, and you can leave the page while it works. Watermarking is switched off for every request, so clips come back clean.
Try Wan 2.7 Text to Video
Two seconds at 720p costs 8 credits and tells you whether the prompt is worth a full 1080p render.